P&ID extraction for data centers.
Data center drawings arrive as two stacked networks on the same set: mechanical P&IDs for the cooling plant (chillers, CRAH/CRAC, CDU, chilled water loops) and electrical one-lines for the power train (utility feed, ATS, UPS, PDU, RPP), both terminating in the BMS and the EPMS. A large share of the points are BACnet or Modbus objects rather than hardwired 4-20 mA loops, so a tag can be a network object as much as a field instrument. Redundancy multiplies the count: an N+1 or 2N facility carries the same measurement on an A-side and a B-side, and each leg is its own point. The sheets are dense with power meters, rack inlet sensors, and leak detection zones that all report back to the same head end.
What data centers drawings carry.
- The same sheet set carries two parallel networks: a mechanical cooling P&ID and an electrical/EPMS one-line. Points from both converge in the BMS head end, so an instrument index has to span field transmitters and power meters together rather than one discipline at a time.
- Many points are not analog loops. They are BACnet objects (Analog Input/Value, Binary Input/Value) per ASHRAE 135, or Modbus registers, so a points list needs a protocol and object-type column alongside the AI/AO/DI/DO signal class, not instead of it.
- Redundancy topology (N+1, 2N, 2(N+1)) drives tag multiplicity. A measurement exists on the A train and the B train as two distinct points; collapsing the pair to one silently deletes a redundancy leg, which matters directly to the Uptime Tier objective.
- Liquid cooling adds CDU-level instrumentation the air-only plants never had: coolant supply and return temperature, coolant flow, conductivity, filter differential pressure, and dew point margin. ASHRAE TC 9.9 W-classes govern the facility water supply temperature these loops run to.
- Leak detection is its own addressable network, not a single instrument. Distributed rope/cable sensor follows chilled water piping and CDU manifolds, spot sensors sit under CRAH units, and zone controllers map physical zones to points, so it reads as zones plus devices rather than one tag.
- Environmental and psychrometric monitoring is dense and interlocked: rack inlet temperature, cold/hot aisle RH, and dew point are tracked against the ASHRAE TC 9.9 thermal envelope. Dew point margin is often an interlock guarding against condensation on chilled surfaces, not just a readout.
- Power quality metering is at almost every node. Utility service, UPS input/output, PDU, RPP, and busway taps each carry a multifunction meter reporting V, A, kW, kWh, power factor and THD to the EPMS. These meters are the raw inputs to the PUE calculation and to the ASHRAE 90.4 MLC/ELC picture.
- Standby power support systems ride the same EPMS: standby generators, day tanks, and fuel oil polishing skids carry their own level, pressure, and temperature instrumentation that has to appear in the index even though it is off to the side of the main cooling P&ID.
- Sequences of operation are tightly coupled to the drawings. Lead/lag pump staging, chilled water valve control, and cooling-loss and fire (NFPA 75) cause-and-effect logic reference the same tags, so a cause-effect matrix commonly sits next to the P&ID as a companion register.
What the extraction delivers.
- BMS/BAS points list (I/O list) carrying signal class plus protocol and object type for networked points
- Instrument and meter index spanning both the mechanical P&IDs and the EPMS one-lines
- Chilled water line list with service, size, and design conditions
- Valve and damper schedule for CRAH/CRAC coils, chilled water branches, and economizer dampers
- Leak detection zone and device schedule (rope/cable runs, spot sensors, zone controllers)
- Power metering schedule feeding the EPMS and the PUE calculation
Instrument tags on a data centers drawing.
- TT-CRAH-12ATemperature transmitter on CRAH unit 12, A-side train. The trailing A/B letter is the redundancy leg, so the paired B-side point is a separate tag.
- FT-CHWS-03Flow transmitter on chilled water supply, loop 3. CHWS/CHWR distinguishes the supply and return headers.
- PDT-COIL-201Differential pressure transmitter across a cooling coil or filter, item 201. Widely used for dirty-filter and coil-fouling alarms.
- MFM-PDU-1AMultifunction power meter on PDU 1, A-feed. Reports volts, amps, kW, kWh, power factor to the EPMS as network points.
- LD-ZN-07Leak detection zone 7, referring to a mapped physical zone on the addressable leak network rather than a single field instrument.
- TT-CDU-05-RETCoolant return temperature on Coolant Distribution Unit 5. RET/SUP mark the return and supply legs of the liquid cooling loop.
Regulatory context.
- ANSI/TIA-942 Telecommunications Infrastructure Standard for Data Centers: governs site, architectural, electrical, mechanical, and fire-safety infrastructure and underpins the redundancy/tier grading.
- ASHRAE 90.4 Energy Standard for Data Centers: sets mechanical (MLC) and electrical (ELC) load-component efficiency limits that shape how densely the plant is metered.
- ASHRAE TC 9.9 Thermal Guidelines for Data Processing Environments: define the air-cooling A-classes and liquid-cooling W-classes that set environmental setpoints and CDU supply temperatures.
- ASHRAE 135 (BACnet): the building automation networking protocol most BMS points ride on, which is why so many tags are network objects.
- NFPA 75 (Protection of Information Technology Equipment) and NFPA 76 (Telecommunications Facilities): drive detection, suppression, and the cooling/fire interlocks tied into the controls.
Which discipline reads it.
HVAC and MEP EngineersThe primary reader for the cooling plant: CRAH/CRAC coils, chilled water loops, CDUs, economizer dampers, and the psychrometric interlocks against the ASHRAE TC 9.9 envelope. Their deliverable is the mechanical points list plus valve/damper schedule.Electrical EngineersOwns the power train one-lines: utility feed, ATS, UPS, PDU, RPP, and the multifunction meters feeding the EPMS. Their pull is the meter schedule and how power points join the same index as the mechanical instruments.Controls EngineersLives at the BMS/EPMS head end where BACnet and Modbus points converge. Cares that the points list carries protocol and object type, that A/B trains stay distinct, and that the cause-effect matrix references the right tags.Commissioning and Startup EngineersUses the index as the checklist for point-to-point verification and IST across redundant trains. Redundancy multiplicity and leak-zone completeness are exactly what they sign off against the Tier objective.
Common questions
Most of our points are BACnet or Modbus objects, not 4-20 mA loops. Does an instrument index even apply here?
Yes. A point is a point whether it is a hardwired loop or a network object. The register keeps the AI/AO/DI/DO signal class where it applies and carries the protocol and object type (for example BACnet Analog Input, Modbus holding register) for the networked points, so both live in one index. You receive a points list where each row identifies how the point is presented, not only what it measures.
We run 2N. If the A-side and B-side carry the same measurement, will the duplicate tags get merged and hide a redundancy leg?
They should not be merged. In a redundant facility the A and B measurements are physically separate instruments on separate trains, and each is its own point. The train suffix on the tag is what keeps them distinct in the index, so both legs stay accounted for and the redundancy is auditable.
Half our sheets are electrical one-lines with power meters and half are mechanical P&IDs. Is the power side covered?
Yes. The multifunction meters on the UPS, PDU, RPP, and busway one-lines are tagged instruments the same as a field transmitter, and they belong in the same index because they report to the EPMS and feed the PUE calculation. Both disciplines land in one instrument and meter index rather than two disconnected lists.
Liquid cooling and CDUs are new on our drawings. Are those instruments recognized?
The coolant supply and return temperatures, coolant flow, conductivity, filter differential pressure, and dew point margin around a CDU are read as ordinary loops and land in the register with the rest of the plant. The facility water supply class these loops run to follows the ASHRAE TC 9.9 W-class the drawing is designed against.
Where does the sequence of operations and the cooling-loss cause-effect fit? We do not want it flattened into the I/O list.
It stays a companion register. The cause-and-effect matrix and SOO reference the same tags as the P&ID, so they are extracted as their own schedule alongside the points list rather than folded into it. The I/O list is the point inventory; the matrix is the logic that acts on those points.
Will any of our topology or design intent be exposed anywhere public?
No. Extracted tags, registers, and drawings stay inside your workspace. Nothing from a customer drawing is used as an example, sample, or reference on any public surface.
Run it on your data centers drawings.
Upload a P&ID set and see the instruments, equipment, and lines it reads before committing to a plan.