In this article7 sections
Two documents in an instrument package have almost the same name, sit next to each other in the deliverable register, and are routinely treated as one thing. They are not one thing. The instrument process data sheet states what the process demands of the instrument. The instrument data sheet states which instrument was chosen to meet that demand. Process engineering owns the first, instrumentation owns the second, and the second is built against the first.
Getting the two confused is not a filing problem. It changes what a vendor quotes against, what a reviewer is signing, and whether a range change three months into detailed design reaches the device that was already ordered.
In one line. The process data sheet is the duty. The instrument data sheet is the device. The duty is issued first, the device is selected against it, and each keeps its own owner and its own revision.
The two documents, side by side
| Instrument process data sheet | Instrument data sheet | |
|---|---|---|
| Issued by | Process engineering | Instrumentation and control engineering |
| Answers | What conditions must this instrument work in | Which instrument, built how, set to what |
| Contains a manufacturer | No | Yes |
| Issued | Early, and revised as the process design settles | After the process sheet it is built against |
| Reviewed by | Process, and the discipline that will select the device | Instrumentation, procurement, and often the vendor |
| Feeds | The technical basis of the requisition and the inquiry | The purchase order, take-off, calibration record, loop folder |
| Invalidated by | A change in the process case | A change in the selected model, or a process change upstream of it |
The direction of that last row is the part worth holding on to. A revision to the process data sheet does not merely annotate the instrument data sheet. It puts the selection recorded on it back in question, because the selection was an answer to the previous duty.
What sits on each
Field lists vary by project and by device type. The split below is the one that stays stable across them, because it follows ownership rather than form layout.
| Field | Process data sheet | Instrument data sheet |
|---|---|---|
| Tag number | Yes | Yes |
| Service description | Yes | Yes, carried forward |
| Line or equipment number | Yes | Carried forward |
| Fluid, phase, composition | Yes | No |
| Flow: minimum, normal, maximum | Yes | No |
| Operating and design pressure | Yes | Referenced for the rating |
| Operating and design temperature | Yes | Referenced for the rating |
| Density, viscosity, molecular weight | Yes | No |
| Vapour pressure, flashing or cavitating service | Yes | No |
| Corrosive or erosive service noted | Yes | No |
| Required range and turndown | Yes | No |
| Allowable pressure drop | Yes | No |
| Line size, schedule, insulation | Yes | Referenced |
| Manufacturer and model | No | Yes |
| Wetted materials of construction | No | Yes |
| Process connection type, size, rating, facing | No | Yes |
| Output signal and protocol | No | Yes |
| Supply power | No | Yes |
| Enclosure rating and area classification | No | Yes |
| Calibrated range | No | Yes |
| Accessories, mounting, manifold, brackets | No | Yes |
| Certification and documentation required | Stated as a requirement | Stated as supplied |
Two rows in that table are where most of the real confusion lives.
Materials. The process data sheet says the service is corrosive, and states what the fluid is and at what concentration and temperature. It does not name the alloy. The instrument data sheet names the wetted material that was selected. If the alloy appears on the process sheet, either process engineering has made a selection decision that is not theirs to make, or the two sheets have been merged and nobody can tell which discipline made the call.
Range. The process data sheet gives the range and turndown the process requires, derived from the minimum, normal and maximum cases. The instrument data sheet gives the range the device is calibrated to, chosen from what the model actually supports. These are two different numbers with a relationship between them, and every calibration record downstream keys off the second one. A project that carries only one range field has lost the ability to show that the calibrated range still covers the required one.
Why they are kept apart
The separation is not administrative tidiness. Four practical consequences follow from it.
Different disciplines sign different things. A process engineer signs that the stated conditions are the design conditions. An instrument engineer signs that the selected device meets them. One signature cannot carry both claims, and a merged sheet asks it to.
Different revision cadence. Process conditions move during process design and then settle. Selections move during procurement, when a vendor offers an alternative, when a material is unavailable, when a delivery date forces a substitution. Putting both on one revision code means every procurement substitution reads as a process change and every process change reads as a purchasing event.
The inquiry needs the duty alone. When an inquiry goes out, what a bidder should be quoting against is the duty. Send a sheet with a manufacturer and model already on it and you have either specified a sole source or invited a bid that simply copies your own selection back at you. Neither is an evaluation.
The as-built record needs the device alone. Calibration certificates, spare parts records and the loop folder all need the instrument that is physically installed. They do not need the flashing calculation that led to it, and a record set that carries process cases into the maintenance system is carrying data that will never be revised again and will be wrong within a turnaround or two.
Where it goes wrong in practice
These are the recurring failure patterns, in rough order of how often they cause rework.
The process sheet is issued late, or as a placeholder. Instrument selection starts against provisional conditions, requisitions go out, and then the process case changes. The resulting rework is usually blamed on procurement. It is a sequencing failure.
A process revision does not propagate. The process data sheet goes to a new revision. The instrument data sheet built against the old one is never revisited, because nothing links the two except the tag number and a convention. The device arrives correct against a superseded duty.
One sheet, one revision code. The most common merge. It works until the first change, at which point the document history stops distinguishing a change in what the process needs from a change in what was bought.
Range fields collapsed into one. The required range is overwritten with the calibrated range once the device is selected, and the requirement it was chosen to satisfy is no longer recorded anywhere.
Control valves treated like transmitters. A control valve carries a process data sheet, a valve specification sheet and a sizing calculation, and the calculation is a third document with its own basis. Folding the calculation into the specification sheet loses the assumptions behind the selected trim and characteristic.
The index and the sheets disagree. The instrument index is the register of which sheets exist and at what revision. When it tracks only one data sheet per tag, the project has no register-level view of whether the duty and the selection are at compatible revisions.
How the pair sits in the wider document set
Reading upward from the P&ID, the chain for a single measuring instrument is usually:
- The P&ID establishes that the measurement exists, its tag and its location in the process.
- The line list and equipment data fix the conditions in that part of the process.
- The instrument process data sheet states the duty for that tag.
- The instrument data sheet records the device selected against that duty.
- The material requisition packages the technical basis for purchase.
- The installation detail fixes how the selected device is physically mounted and connected.
- The loop diagram and the calibration record close out on the device as installed.
Steps 3 and 4 are the hinge. Everything above them is process definition and everything below them is execution against a chosen device, which is exactly why the two are separate documents with separate owners.
What to check on a package you have been handed
A short review pass that catches most of the above:
- Does every tag with a data sheet also have a process data sheet, and are both listed in the index at a revision?
- Does the instrument data sheet cite the process data sheet revision it was built against?
- Is the required range recorded separately from the calibrated range?
- Does any process data sheet name a manufacturer, a model or an alloy? If so, ask who made that decision.
- For every process data sheet revised since the instrument data sheet was issued, has the selection been re-confirmed?
- For control valves, is the sizing calculation present as its own document with its stated cases?
None of that requires opening a vendor catalogue. It is a register-level check, and it is the cheapest point at which to catch a duty and a device that have drifted apart.