The EPMS power-meter points list is the register of every metering point on a data center's electrical one-lines, from the utility service entrance down through the UPS, PDU, and RPP layers. It is the electrical counterpart to the mechanical points list, and the document that feeds the PUE calculation.
This piece walks the power side of the drawing set: how to read metering off the one-lines by distribution level, how to keep A-side and B-side meters distinct in a redundant facility, and what the register needs before it is complete enough to compute Power Usage Effectiveness.
This is a working reference for the electrical and controls engineer building that register, not a marketing note. If you spend your days reading UPS and PDU one-lines and reconciling meter tags against an EPMS point map, this is the structure the register wants to take.
Key takeaways
- The power-meter register spans the electrical one-lines the way an instrument index spans the mechanical P&IDs: utility, generator, ATS, UPS, PDU, RPP, and busway metering all in one list.
- PUE is total facility energy over IT equipment energy. The register must identify which meters feed the numerator and which feed the denominator, because meter placement, not arithmetic, is what makes the ratio right.
- Redundancy multiplies the count. In N+1 or 2N, the A train and B train each carry their own meters, and each leg is a separate row keyed by its train suffix.
- Many meters report over Modbus or BACnet rather than a hardwired loop, so the register carries a protocol and register or object reference alongside the electrical quantities.
- A complete register names the distribution level, the redundancy leg, the parent bus, and the reported quantities for every meter, so the PUE roll-up is auditable rather than assumed.
Two networks on one sheet set
A data center drawing package carries two parallel networks. The mechanical P&IDs describe the cooling plant: chillers, CRAH and CRAC units, chilled water loops, and increasingly the CDU-level instrumentation that liquid cooling adds. The electrical one-lines describe the power train: the utility feed, the standby generators and ATS units, the UPS, the PDU, the RPP, and the busway that reaches the racks. Both networks terminate in monitoring head ends, the BMS for the mechanical side and the EPMS for the electrical side, and a large share of the points on both are network objects rather than hardwired 4-20 mA loops.
The cooling side and its BMS points list are a separate exercise, covered on the building automation side of the cluster. This piece stays on the power train and the EPMS. The two registers share a drawing package but answer different questions, and keeping them scoped apart is what stops the electrical meter schedule from drowning in psychrometric points that belong to the cooling plant.
A data center drawing set is two networks in one binder. Keep the power meters and the cooling points in separate registers, or each one drowns the other.
Metering by distribution level
The power one-line is a hierarchy, and the metering follows it. Reading the register level by level is how you keep the count honest, because every distribution node between the utility and the rack tends to carry its own multifunction meter reporting to the EPMS. A meter at each level answers a different question: how much the building draws, how much survives the UPS, how much reaches a given PDU, and how much lands on a branch.
The table below walks the levels from the service entrance down, with synthetic tags in the style the sheets tend to use.
| Distribution level | Example meter tag | What it carries |
|---|---|---|
| Utility service entrance | MFM-UTIL-01 | Total facility power at the main switchgear. Natural home of the PUE numerator, before any distribution split. |
| Standby generator | MFM-GEN-03 | Generator output during a utility event. Sits off the main one-line but rides the same EPMS. |
| ATS / bus tie | MFM-ATS-02A | Which source is feeding the bus, and the load carried across the transfer. |
| UPS input and output | MFM-UPS-2A, MFM-UPS-2B | Load through each UPS module. The A and B modules are separate points on separate trains. |
| PDU | MFM-PDU-1A | Volts, amps, kW, kWh, and power factor at the PDU. Often the boundary where IT energy is metered for the PUE denominator. |
| RPP / branch | MFM-RPP-14A | Load on a remote power panel feeding a row or a containment pod. |
| Busway tap | MFM-BUS-07A | Load at a busway tap-off to a rack or a small group of racks. |
The MFM stem, for multifunction meter, is a convention rather than a rule; some drawing sets use PM, PQM, or a manufacturer-specific prefix. What matters is that the register captures the level, not that every tag shares a prefix. The register is what makes the hierarchy legible when the tags do not.
The A/B suffix is a distribution fact, not a label
The trailing A and B on MFM-UPS-2A and MFM-UPS-2B is the redundancy leg. In a 2N facility the A train and the B train are two complete, independent power paths, and the same measurement exists once on each. Those are two physically separate meters on two separate buses, and they are two rows in the register. The suffix is what keeps them distinct.
This matters directly to the Uptime Tier the facility was designed to. A Tier III or Tier IV grade is a claim about concurrent maintainability and fault tolerance, and it rests on the redundancy being real and auditable. A register that quietly merges MFM-PDU-1A and MFM-PDU-1B into a single PDU meter has erased the evidence that the second path exists. The suffix carries the topology; keep it intact through every downstream export.
The A and B suffix is not a label, it is the topology. Merge the two legs into one meter and you have erased the proof the redundant power path exists.
What PUE needs from the register
Power Usage Effectiveness is a simple ratio with a demanding measurement boundary. It is total facility energy divided by the energy delivered to the IT equipment. A value near 1.0 means almost everything the building draws reaches the IT load; a higher value means more of the draw is going to cooling, distribution losses, and support systems. The arithmetic is trivial. The difficulty is entirely in which meters you read for the numerator and which for the denominator.
The numerator, total facility energy, is naturally the utility service entrance meter, because it sees everything before any split into IT and non-IT paths. When a site has more than one service entrance, or on-site generation feeding the same bus, the numerator is the sum of those metered feeds, and the register has to mark which meters aggregate into that sum.
The denominator, IT equipment energy, is measured at the boundary of the IT load. The common boundaries are the UPS output or the PDU output feeding the white space. Where you place that boundary changes the ratio, which is why the measurement category matters and why the register cannot be silent about it. A PDU-output meter and a UPS-output meter give different denominators for the same facility, and the register is the document that records which boundary this PUE was computed against.
So a register that is going to feed a defensible PUE needs, per meter, at minimum:
- The meter tag and its distribution level.
- The parent bus or upstream node it hangs off, so the roll-up tree is explicit.
- The redundancy leg, so A and B are summed correctly rather than double-counted or dropped.
- A flag for whether the meter contributes to the numerator, the denominator, or neither.
- The reported quantities and the protocol or register reference, so the EPMS integrator knows what to poll.
Without the numerator and denominator flags, the register is a meter inventory but not a PUE input. The flags are the part that turns a list of meters into a calculation.
Register columns for the power side
An EPMS meter register borrows the spine of an instrument index and adds the electrical distribution context. A typical column layout:
| Column | Purpose |
|---|---|
| Meter tag | Master identifier, e.g. MFM-RPP-14A. |
| Distribution level | Utility, generator, ATS, UPS, PDU, RPP, busway. |
| Parent node | The upstream bus or panel this meter hangs off. |
| Redundancy leg | A, B, or common, from the tag suffix. |
| Reported quantities | V, A, kW, kWh, power factor, THD as applicable. |
| Protocol / reference | Modbus register map or BACnet object, for networked meters. |
| PUE role | Numerator, denominator, or none. |
| Service description | What this meter measures, in plain words. |
That column split lets you filter the register by level, sum a redundancy pair correctly, and trace the PUE roll-up without parsing tag strings by hand. It is the same discipline that a good instrument index applies to field transmitters, applied to the power train. If you are new to the underlying document, the I/O list reference explains the signal-class columns that sit alongside these electrical columns for the hardwired points.
Networked meters need a protocol column
Most modern power meters report over a network rather than a hardwired analog loop. A multifunction meter on a PDU typically presents its volts, amps, kW, kWh, and power factor as Modbus registers or as network objects the EPMS polls. That does not change what the meter is, but it changes what the register has to record about it.
For a networked meter, the AI/AO/DI/DO signal class that describes a hardwired loop does not fully apply, because a single meter presents many measured values over one connection. The register carries a protocol and reference column so the EPMS integrator knows the map: which register holds kWh, which holds power factor, which holds the alarm status. A hardwired current transformer feeding a discrete input and a networked meter reporting fifty values both belong in one register, described by what they measure and where they sit, with the communication method as an attribute rather than a separate list.
Standby power belongs in the register
It is tempting to treat the register as ending at the UPS, since that is where the clean, conditioned power to the IT load begins. The standby side is part of the same EPMS and part of the same PUE story. Standby generators, ATS units, day tanks, and fuel polishing skids carry their own metering and their own level, pressure, and temperature instrumentation, and they report to the same monitoring head end. During a utility event they are the facility's power source, so their metering is not an afterthought; it is the record of whether the site rode through the event on its own generation.
Including the standby metering keeps the register aligned with how the facility is actually operated and graded. The ANSI/TIA-942 infrastructure grading and the Uptime Tier both assume the standby path is instrumented and monitored, so a register that stops at the UPS is describing only the fair-weather facility, not the one the Tier claim was made about.
What "complete" means for this register
A power-meter register is complete when it can produce the PUE numerator and denominator without a side conversation. Concretely, that means every metered node on the one-lines appears once, every A/B pair is present as two rows keyed by leg, every meter names its parent node so the roll-up tree is explicit, and the numerator and denominator flags are set. At that point the register is not just an inventory of meters; it is a calculation waiting for energy values, and any engineer can trace how a reported PUE was assembled.
Pulled off a full drawing set, that register comes back as one power-metering schedule with every meter, its distribution level, and its redundancy leg accounted for, ready to feed the PUE roll-up and the ASHRAE 90.4 electrical load-component picture. From there it is the electrical engineer's document to review, correct against the one-line, and hand to the EPMS integrator.
If you are building the cooling-side companion to this register, the data center hub collects the mechanical points list, the leak-detection schedule, and the rest of the cluster alongside this power-side piece.