Units of Measure

Infohub does not convert units#

The unit is stored on the measure as free text. There is no unit catalogue, no dimensional analysis, and no automatic conversion.

That means:

• Infohub will let you divide litres by hours and label the result kg.
• Infohub will let you aggregate a measure in kW into one labelled kWh.
• Infohub will let you write m3, m³, M3, and cubic metres on four measures holding the same quantity, and treat them as unrelated labels.
• Nothing will warn you about any of it.

The unit is documentation for humans and a label on charts. Its correctness is entirely your responsibility, which makes convention the only defence.

Agree a convention and write it down#

Because Infohub will not enforce anything, agree the spellings once, record them somewhere your team will find, and apply them everywhere. The specific choices matter less than the consistency.

A workable starting set:

Two practical notes on character choice. Superscripts and degree signs (m³, °C) render correctly in the web client but can be mangled by downstream systems, CSV exports, and some databases — if data leaves Infohub through a consumer, plain ASCII (m3, degC) is the safer choice. And keep case consistent: KG, Kg, and kg are three different strings to every grouping and search you will ever run.

Power is not energy#

The single most common unit error in plant reporting.

Power is an instantaneous rate — kW. Energy is power accumulated over time — kWh. They are different quantities and cannot share a unit.

The correct chain is unforgiving about which aggregation you use:

Power                Tag      Retrieved      kW    instantaneous reading
+-- Power Hourly     Hourly   Average        kW    average power in the hour
    +-- Peak Demand Monthly  Monthly  Maximum  kW    highest hourly average

Energy Meter         Tag      Retrieved      kWh   cumulative register
+-- Energy Hourly    Hourly   Counter        kWh   energy used this hour
    +-- Energy Daily Daily    Sum            kWh   energy used today

Averaging a power reading gives you power. Summing power readings gives you nothing meaningful. If a measure labelled kWh was built with an Average aggregation from a kW tag, it is wrong — see Choosing the Right Aggregation Type.

Units on calculated measures#

The result unit of a calculation is not derived — you set it, and you should set it from the arithmetic rather than from habit.

For percentages, be certain whether the expression multiplied by 100. A ratio of 0.924 labelled % will be read as 0.924%, and a value of 92.4 formatted with 0.0% displays as 9240%. See Format Strings.

Where to convert#

When a source reports in a unit you do not want, convert deliberately and visibly.

Best: fix it at the source. Correcting a historian that logs in the wrong unit fixes it for every consumer, not just Infohub.

Good: a calculated measure. Retrieve raw in the source's unit, then a calculation converts to yours. The conversion is explicit, reviewable, and the raw value stays visible for auditing.

Acceptable: in the source query. SQL-based connectors can scale in the query. Concise, but the conversion is buried in a connection property where nobody will look for it.

Never: relabelling. Changing the unit text without changing the value is not a conversion. It is a silent, permanent error that every downstream reader will trust.

For repeated conversions, put them in a custom function with an unambiguous name — KgToTonnes, not Convert.

Mixed-unit sites#

Sites that report in both metric and imperial, or that report locally in one system and to a parent company in another, are common.

Do not try to make one measure serve both. Build the model in one canonical system, then add converted measures where the second system is genuinely needed, named so the unit is obvious:

Production Daily          t     canonical
Production Daily (short tons)  ton   converted, for group reporting

Explicit duplication is far safer than a measure whose unit depends on who is looking at it.

Auditing units in an existing model#

Units drift as a model grows and different people add to it. A periodic check is cheap:

• List every distinct unit string in the model. Near-duplicates — m3 and m³, t and ton, degC and °C — show up immediately.
• Check every measure whose name contains "per", "rate", "%", "efficiency", or "specific": these are all divisions, and their units are the most frequently wrong.
• Check for blank units.
• Check that anything labelled kWh was built with Counter and Sum, and anything labelled kW with Average or Maximum.

The Config CLI can list measures with their units, which makes this a scriptable check rather than a manual trawl.

See also#

• Format Strings
• Calculated Measures
• Retrieved Measures
• Folder Structure Best Practices