Armeta's contextualization engine reads any PDF Process Flow Diagram — scanned legacy drawing, rasterized CAD plot, flat image export, vendor deliverable — and extracts structured process data: equipment blocks, stream connections, flow rates, temperatures, pressures, compositions, and material balances. The output is contextualized against your P&ID and line list data, so high-level process intent connects to the detailed engineering record.
The PFD is the document that explains what the process does. Armeta extracts every entity on the sheet so that intent becomes queryable rather than locked in a flat image.
Vessels, columns, exchangers, drums, compressors, pumps, packages — every tagged equipment block on the PFD identified, classified, and resolved against your tag register.
Every stream between equipment, with stream number, source, destination, and direction parsed into structured connectivity ready for graph assembly.
Mass and volumetric flow rates per stream extracted with units normalized so the structured output is queryable across vintages and authoring conventions.
Stream temperatures captured per stream and reconciled against design and operating columns on the line list where present.
Stream pressures captured per stream, with units normalized and any operating-vs-design distinctions preserved.
Stream compositions parsed from the stream-data table, with components normalized against your composition register where one exists.
Mass and energy balance data — whether shown on the PFD itself or referenced as a separate HMB — structured per stream and tied to the source.
Off-PFD references and battery-limit connections to other units captured as structured edges so the graph spans facility-wide rather than stopping at the unit boundary.
The PFD is the highest-level engineering document of the process. The P&ID and the line list are the granular detail. Armeta resolves them into a single engineering knowledge graph so design intent and engineering record are addressable as one model.
Equipment tags on the PFD resolve to their detailed representations on the P&ID, and stream data reconciles against line list attributes — with discrepancies surfaced for engineering review.
Every equipment block on the PFD is matched to its detailed depiction on the P&ID. The PFD-level vessel becomes the same entity as the P&ID's vessel, with all of the P&ID's nozzles, instruments, and trim surfaced under it.
PFD streams are reconciled against the line list. Where a PFD stream corresponds to one or more line-list lines, attributes — size, spec, service, design conditions — are cross-checked. Mismatches are surfaced as discrepancies, not silently absorbed.
Operating conditions on the PFD are compared against design and operating values on the line list and against pressure/temperature annotations on the P&ID. The graph keeps each value with its source so engineers reconcile the differences with full evidence.
The PFD and the P&ID are not interchangeable. They describe the same process at different levels of detail, and each captures what the other deliberately leaves out. Both are needed to understand the unit — and both belong in the engineering knowledge graph.
Major equipment, principal streams, and how the process flows at the unit level. The document an engineer reads to answer “what does this process do?”
Mass and energy flows per stream. The quantitative backbone of the process design — typically absent from the P&ID.
Component breakdowns per stream. The basis for utility loading, environmental compliance, and downstream specification.
Block valves, control valves, check valves, isolation valves — every device is on the P&ID. The PFD shows none of them.
Local drains, vents, sample points, and minor utility connections live only on the P&ID. The PFD aggregates them away.
Equipment, line, and instrument tags resolved at full granularity. The PFD names equipment but does not tag every component beneath it.
Once the PFD is structured and contextualized against the P&ID and line list, system-level engineering decisions stop being driven by re-reading flat images.
Confirm the PFD against the heat and material balance and against the developing P&IDs during detailed engineering — without re-keying stream data into a spreadsheet.
Reconcile the heat and material balance against the structured PFD streams to surface unit-level imbalances early, when they are still inexpensive to fix.
Every PFD extraction Armeta delivers is validated against the source drawing by our engineering team before it reaches you. Accuracy metrics on your own PFDs are produced as part of the initial document audit — our ten-document free assessment that establishes baseline performance before any paid engagement.
Every PFD extraction is reviewed by an Armeta engineer against the source drawing before delivery. Not raw model output.
Every equipment block, stream, and operating-condition value links back to the exact region on the source PFD it was extracted from.
Stream-data tables are cross-checked against on-PFD annotations and against any referenced heat and material balance — discrepancies surfaced rather than averaged.
Your stream-data schema is preserved. Columns are not dropped, not renamed, and not silently merged — the structured output round-trips cleanly back to your master.
Three deployment modes, chosen to match the security and governance constraints your engineering records actually live under.
Fastest time to value
Armeta-hosted, isolated per customer. Your engineering documents and extracted data remain segregated from every other customer environment.
Your network, your perimeter
Armeta's contextualization engine runs inside your network. Your documents never leave your environment. Deployed with customers today under data residency and air-gap requirements.
Managed inside your tenant
Armeta-managed deployment inside your own cloud tenant. Cloud operational model, customer-controlled perimeter.
Before any procurement conversation, see Armeta run on your actual PFDs — equipment, streams, and stream data resolved against your P&IDs and line lists. Not sample data. Not a generic demo. Your PFDs, contextualized, delivered as a structured report in one engagement.
