From DOE to P&ID: where engineering handoffs lose information

A process usually reaches its P&ID after passing through four or five separate tools: a statistics package for the design of experiments, a simulator, a drafting tool for the PFD, another for the P&ID, and a set of spreadsheets holding everything in between.

Each tool is good at its job. The trouble is in the gaps between them.

The typical chain

1. DOE. Trials are planned and run; a response model is fitted.
2. Simulation. Results become the basis of a process model, which is solved for heat and material balances.
3. PFD. Streams and equipment are drawn, and balance data is copied into a stream table.
4. P&ID. Piping, valves and instruments are added, with tag numbers and line specifications.
5. Design. Equipment is sized and datasheets are written.

At every arrow, a person reads values from one system and types them into the next.

Where information gets lost

Re-keyed numbers. A flow rate copied by hand from a simulation report is correct until someone mistypes it. It is also correct only until the simulation changes.

Unit conversions. Values move between tools with different unit sets. A conversion done once in a spreadsheet can quietly be done twice, or not at all.

Tag drift. Equipment called E-101 on the PFD becomes E-1001 on the P&ID after a numbering scheme changes. Nothing breaks until someone tries to match a datasheet.

Stale revisions. The P&ID is updated, but the stream table behind it came from last month's simulation case.

Assumptions left behind. The basis of design often lives in an engineer's notes. The number travels to the next stage; the reason for it does not.

Why it matters

None of these is dramatic on its own. Together they create review cycles spent finding transcription errors instead of engineering ones, and they push discovery late: into a HAZOP, into procurement, or onto site, where every correction costs more.

A continuous model instead of copies

The alternative is to let each stage reference the one before it rather than copy it. The PFD reads its stream data from the converged simulation. The P&ID inherits equipment and tags from the PFD. Datasheets draw on both.

When an upstream value changes, everything downstream that depends on it is known, and can be updated or flagged for review. The engineer still decides what to accept. What disappears is the silent drift.

This is the idea Ardsol is built around: five stages, one continuous model, with a verification layer checking each handoff before work moves on. See how it works.