Free product samples available for qualified projects — Request yours today
Home/Blog/Article

When to Replace a Smart Energy Meter — And Why Most Factories Get It Wrong

Over the past four years, I've reviewed more than 200 unique energy monitoring specs, acceptance reports, calibration certificates, and communication logs across our production sites. At GridSense, I'm the person who signs off—or refuses to—before an installation gets accepted.

And I'll say it directly:

Most factories don't need to replace their smart meters. They need a better submetering strategy.

I mean that. The "replace all meters" reflex is one of the most expensive misdiagnoses in factory power monitoring, and it almost always shows up on page one of the proposal.

Why "Replace the Meters" Became the Default Answer

Here's the thing—when energy data looks wrong, the meter is the most visible suspect. Missing circuits. Flat-lined trends. Overnight gaps in the trend chart. The meter is the branded box on the panel, so when a plant stops trusting its data, replacing the meter feels like a clean fix.

It's tempting to think a newer meter will automatically produce better data. But a replacement only solves problems that live inside the meter—and most of the problems I audit live somewhere else entirely.

And "replace the meters" is a diagnosis the vendor is usually happy to confirm. They sell meters (surprise, surprise).

I'm not saying meters never fail. I'm saying that on the sites I've audited, the meter is usually the last thing I'd suspect. Usually.

Argument 1: Smart Meters Last Longer Than You Think

In our Q1 2024 quality audit, I pulled five years of calibration certificates from our three main production sites. Not one meter had drifted outside its declared accuracy class under normal operating conditions.

The relevant standard—IEC 62053-22 for Class 0.5S and Class 0.2S static watt-hour meters—defines allowable percentage error under specified operating conditions. It's not a warranty, but it tells you what the hardware is designed to do. Static meters in temperature-controlled panels, running within their rated current, often hold their class for the better part of a decade.

I'm not sure why some vendors push replacement at year five. My best guess is that their maintenance contracts are structured that way, or the meters are the easiest line item to justify when a bigger platform upgrade is on the table.

Either way: before scheduling a replacement, ask for the last calibration records. If the meters are inside their declared class and the records are current, replacing them is paying for the illusion of a fix, not a fix.

Argument 2: The Weak Points Are Never the Meter

Here's the counterintuitive part—the one that keeps catching buyers off guard.

Most factories focus on the meter's accuracy class and completely miss the two components that determine what the platform actually receives: the CT (current transformer) and the communication gateway.

Take CT sizing. I've reviewed installations where a 400/5A CT was paired to a branch circuit that normally carries 60–80A. At low load, the CT output is tiny, and small absolute measurement errors translate into large percentage errors in the reported energy. The meter is fine. The data is garbage.

Then there's the communication layer. I've pulled gateway logs where the actual acquisition interval was 15 minutes, not the 1-minute interval the proposal promised. Nobody caught it because nobody looked. The "smart" meter was reporting correctly; the platform was storing a smoothed, delayed picture that made every load spike invisible.

One more thing on CTs: accuracy class has to match, too. Pairing a Class 1.0 CT with a Class 0.5S meter doesn't give you 0.5S accuracy—it gives you 1.0-class results with 0.5S-class paperwork. That mismatch is depressingly hard to spot after commissioning.

For any smart submetering solution for factories—including the power monitoring work we do at GridSense—the platform-facing question isn't "what accuracy class is the meter?" It's "what does the platform actually see, and how often?" Those are different questions, and only one of them is about the meter.

I still kick myself over one site where we almost approved a full meter replacement because of a 15-minute gateway interval that made trends look wrong. Once the logs were pulled, the picture was obvious. That budget should have gone to twelve panels that had never been metered at all.

Argument 3: When Replacement Actually Makes Sense

That said, there are legitimate cases. I've signed off on replacements myself. Here's what pushes me to endorse one:

  1. Confirmed accuracy drift. Not assumed, not "the vendor says it's old." Confirmed against a calibrated reference, outside the declared class. That's a fact, not a hunch.
  2. Protocol obsolescence with no viable bridge. If the meter speaks a legacy protocol your platform genuinely cannot consume, and no supported gateway bridges it, replacement is defensible. "Legacy" alone isn't grounds—check for bridges first.
  3. End of manufacturer calibration support. If the vendor has stopped calibration services and firmware updates for your model, and you need regulatory-grade readings, that's a real constraint.
  4. Regulatory or settlement requirements. If the utility or grid operator requires a certified meter for settlement, that's not your call.
  5. Known firmware security issues. If your threat model requires patched firmware, and the vendor has stopped supporting the version, replacement becomes a security decision—not a technical one.

Notice what's not on that list: "the vendor released a newer model" and "the proposal includes meters." Neither of those is a reason.

But the Vendor Says It's Outdated

Fair pushback. "Outdated" and "failing" are not the same thing. A meter released in 2017 might not have the newest communication stack, but if it's producing data inside its class and your platform can ingest the readings, it's doing its job.

The other pushback I hear: "But the new meters have more features." Sure—and feature counts don't fix data quality. If you haven't mapped submetering coverage, if the CT ratios don't match actual load, and if the gateway acquisition rate doesn't match what the platform needs, a new meter with a longer feature list will reproduce the same gaps in a more expensive box.

Look, I'm not anti-replacement. I'm anti-replacement-before-diagnosis.

The Order of Operations I Actually Run

If a plant tells me "our energy data looks wrong," here's the sequence I follow before I approve anything:

  1. Map the coverage. Which panels are metered? Which branches aren't? A lot of "bad data" complaints are really "missing data" complaints. Submetering coverage is a strategy question, not a hardware one.
  2. Check CT ratios against actual load. Standard practice is to size CTs so expected load lands in the 40–80% range of the CT's rated primary current. A 400/5A CT on a 60A circuit is the wrong tool. Compare the nameplate to the historical load profile.
  3. Pull the communication logs. What's the actual acquisition interval? Any overnight gaps? Are gateways buffering and dumping? The platform is only as good as what it receives.
  4. Review calibration records. If the meters are inside their class and the records are current, leave them alone. Spend the budget on coverage.
  5. Then—and only then—consider replacement, for one of the documented reasons above.

If a proposal skips the first three steps and jumps straight to "replace all meters," it hasn't diagnosed anything. It's just selling.

Fix the Granularity Before the Hardware

Energy monitoring IoT platforms are only as good as the data they're fed. No platform can recover signal from a CT that's mismatched to its load, a gateway that samples every 15 minutes, or a panel that was never sub-metered in the first place.

So before you budget for new smart meters, budget for a proper submetering audit: coverage map, CT sizing, acquisition intervals, calibration records. The meters are probably fine. The strategy behind them is usually what needs replacing.

That's the part most proposals miss. And it's the part that actually moves your data quality.

author avatar
Sarah Okonkwo

Sarah Okonkwo is an electrical test and measurement analyst specializing in multimeters, clamp meters, installation testers, sensors, and power-quality monitors. She uses IEC 61010-1 and IEC 61010-2-032 safety requirements for test and current-clamp equipment, then applies IEC 61000-4-30 methods to power-quality parameters while examining category ratings, accuracy, resolution, bandwidth, harmonics, voltage events, and uncertainty. She helps technicians and buyers select instruments, set up safe measurements, interpret readings, and diagnose faults within the intended circuit environment.

Leave a Reply

Your email address will not be published. Required fields are marked *

Please write your comment.