If “you can’t manage what you can’t measure,” then electrical meters provide the critical power measurements needed to manage data centers. Specific types of meters serve various purposes, from warning operators of impending power problems (overloaded circuits) to identifying root causes of downtime (power quality) to quantifying the cost of inefficiencies (UPS, transformer, harmonic losses) to reporting power usage effectiveness (PUE). Which type of meter to incorporate into a data center design depends on what needs to measured, and therefore managed.
These objectives dictate whether to capture data at the building, switchboard, circuit, or end-use level—particularly for PUE reporting. For instance, a dedicated data center would meter at the building (total data center power consumption) and end-use (UPS or PDU output as proxy for IT load) levels. But a shared-facility data center would require meters at switchboards and circuits to measure data center-specific power consumption.
Pros and cons of the 7 types of electrical meters
There are two categories of electrical meters: stand-alone devices and meters embedded into other infrastructure equipment.
Stand-alone power quality meters measure electrical parameters (energy, voltage, current, power factor, frequency) and monitor power quality (harmonics, voltage disturbances). They are highly accurate (± 0.2%) and precise (1024 samples per cycle) devices that provide detailed engineering data (e.g., waveforms). Consequently, they are expensive and require an engineer to interpret the data.
Simpler devices with a more targeted data set, power meters measure the same kind of electrical parameters but with a lesser degree of accuracy (± 0.5%) and precision (32 samples per cycle). They are less expensive but offer more limited power quality monitoring.
Although primarily protective devices for medium voltage equipment, digital relays may embed meters into advanced models. Leveraging their built-in metering capability avoids the higher cost of stand-alone meters, but accessing the data is more difficult. Because digital relays are meant for large amperage inflows, their meters do not measure as accurately (± 1%).
Electronic trip unit embedded meters can often replace both power quality meters and power meters within the electrical infrastructure. This metering option is not only less expensive but also easier to manage—electronic trip units are integrated right into the design of low voltage circuit breakers. Disadvantages are less-accurate data, less-than-full power quality metering, and greater difficulty in accessing the data.
Built-in UPSs meters can serve as a proxy for Category 1 PUE reporting. Note, however, that non-IT loads (air handlers, PDUs) powered through the UPS would be inaccurately counted as IT load. Relying on UPS meters has the advantage of eliminating the cost of additional meters closer to the load. Disadvantages are less-accurate data and lack of power quality monitoring.
Power distribution units (PDUs), as well as remote power panels (RPPs) and busways, embed meters to varying degrees to measure Category 2 PUE cost-effectively. They share the same tradeoffs as UPS meters: non-IT power would be calculated as IT, and power quality monitoring is not comprehensive.
Metered rack PDUs capture data for Category 3 PUE reporting most accurately. Measuring power loads on an outlet-by-outlet basis avoids the kind of complicated mapping needed to figure out which breakers feed IT equipment. Metered rack PDUs cost about 50% more than basic PDUs and in general measure data less accurately than the other upstream meters. In addition, using them in only some racks gives an incomplete picture of the IT load.
In sum, embedded meters are cost-effective and should be implemented whenever possible.