Tenant Energy Submetering: Get the Numbers Right Before You Bill

Every submetering dispute I've been pulled into started the same way. A tenant gets a bill that looks wrong, asks how it was calculated, and the property manager discovers nobody can fully explain it. Sometimes a CT was installed backwards. Sometimes a month of data vanished and someone typed in an estimate. Sometimes the bill was correct and the tenant just didn't trust it. In all three cases, the cost of the argument exceeded the value of the energy in question.

Submetering in commercial and mixed-use buildings is simple in principle. Put a meter on each tenant's supply, read it regularly, bill accordingly. The trouble is in the details that get decided quickly during construction and then live with the building for twenty years.

Accuracy Class Is Only Part of the Story

Meters are sold with an accuracy class, and specifications usually call for revenue grade, typically 0.5 or 0.2 class under the relevant ANSI or IEC metering standards. That number describes the meter under defined test conditions. It says nothing about the current transformers feeding it, and in the field the CTs are where most of the error comes from.

CT Sizing Mistakes

The most common mistake is sizing CTs to the breaker rather than to the expected load. A tenant panel on a 400 A feeder might draw 40 to 80 A on a normal day. Put 400 A CTs on it and the meter spends most of its life measuring at 10 to 20 percent of CT rating, and far lower at night and on weekends. CT accuracy degrades at the low end of the range, so a 0.5 class meter paired with oversized CTs can drift into errors of several percent. On a large tenant that is real money, and it tends to push consistently in one direction.

Other mistakes I see regularly:

• CTs installed with the polarity arrow pointing the wrong way, producing negative or nonsensical readings on that phase
• CT phases swapped relative to the voltage references, so phase A current gets multiplied by phase B voltage and the power factor looks impossible
• CT ratios entered incorrectly in the meter setup, often after a CT was swapped during construction
• Voltage taps landed downstream of the tenant's main disconnect, so the meter goes dark whenever that disconnect is opened

Every one of these is caught by a simple commissioning check: compare per-phase current, voltage and power factor against a clamp meter under known load. It takes maybe fifteen minutes per meter. It gets skipped constantly, usually because the electrician has left the job and the controls integrator assumes the meter is fine because it's reporting numbers.

Getting the Data Out: Modbus, M-Bus and Pulse

How the meter communicates determines how much you can do with the data later.

Pulse outputs are the oldest and simplest. Each pulse represents a fixed quantity of energy, and a counter somewhere adds them up. It works, and for some water and gas meters it's the only option. But a pulse only tells you consumption. You lose demand detail unless pulses are timestamped carefully, you get no power quality information, and a missed or doubled pulse during a power blip silently corrupts the total. If the counter resets, you may not notice for weeks.

Modbus, over RTU or TCP, is the workhorse for electric submeters in commercial buildings. You get registers for cumulative kWh, demand, per-phase values and often power factor and harmonics. The catch is that register maps differ between meter models, scaling and word order vary, and an integrator who maps a 32-bit float with the wrong byte order will produce numbers that look plausible until someone checks. Always verify cumulative energy against the meter's own display after mapping.

M-Bus is common in many regions for heat, water and some electric meters, especially in residential and mixed-use buildings. It was designed for exactly this job: many meters on a two-wire bus, read by a central master, with standardized data records. Wired M-Bus is dependable over long runs. Wireless variants make retrofits easier but bring their own gaps when a new partition wall blocks a meter's radio path.

My preference for electric submeters is anything that exposes cumulative energy registers, not just interval values. Cumulative registers let you recover from a communication outage. Interval-only data doesn't.

Data Gaps and Honest Estimation

Gaps will happen. A gateway reboots, a network switch gets replaced during a fit-out, someone unplugs the data logger to charge a phone. You will have gaps. What matters is how you handle them.

If the meter keeps a cumulative register, total energy for the billing period is still accurate even with missing intervals. You just don't know how it was distributed over time. That matters if you bill on time-of-use rates or demand. On a flat energy rate it barely matters at all.

When the cumulative value itself is lost, you have to estimate. Reasonable methods include using the same days from the previous week, a regression against outdoor temperature for HVAC-heavy loads, or the tenant's average for comparable days. What isn't reasonable is filling the gap silently. Flag estimated intervals in the database, show them on the bill, and set a rule for how long a gap can run before the tenant is told. Utilities have worked under estimation rules for a long time, and the principle transfers well: estimates must be documented, repeatable, and corrected when real data comes back.

Turning Interval Data Into a Fair Bill

The bill is where engineering meets the lease, and the lease usually wins. Read it before you design the billing logic. Some leases allow passing through the actual utility tariff, including demand charges and time-of-use rates. Others specify a flat rate per kWh, or a share of common area energy based on floor area.

Demand charges are where fairness gets tricky. The building's peak is set at one moment, and a tenant's contribution to that peak can be very different from its own monthly peak. Allocating demand by each tenant's load at the time of the building peak is fairer than using each tenant's individual peak, but harder to explain on one page. Pick a method, write it down, and apply it consistently.

Common area loads, central plant energy and losses also need a clear allocation. If the submeters add up to 85 percent of the main meter, the remaining 15 percent has to go somewhere, and tenants will ask where. A monthly reconciliation comparing the sum of submeters to the utility meter is the single best check on the whole system. A widening gap usually means a failed meter or an unmetered load someone added without telling anyone.

What Submeter Data Does Beyond Billing

Owners who install submeters only for billing tend to underuse them. The same interval data shows a tenant's base load at 3 a.m., which in offices with poorly scheduled equipment can be a surprisingly large fraction of the daytime load. It shows rooftop units running all weekend, or a server room's cooling load creeping upward month after month. Comparing similar tenants on a per-square-meter basis quickly tells you where to send a technician first.

It helps maintenance too. A sudden shift in power factor or phase imbalance on one panel can point to a failing motor or a capacitor problem before anyone files a comfort complaint. And when a tenant wants to set its own energy targets, submeter data is what turns that conversation from theory into something you can measure.

So before the next submetering spec goes out, check one thing. Did anyone size the CTs from the expected load profile, or did they just copy the breaker rating off the panel schedule?