Why Your Condition Monitoring System Might Be Missing the Real Problem—and What it Costs You

The Problem Everyone Thinks They Have

I've reviewed a lot of monitoring system specifications over the years. As a quality manager, I see the same thing in about 70% of first submissions: a dashboard full of data that looks impressive but doesn't actually answer the core question—is this system operating as designed?

When I started in 2020, I assumed the problem was technical. Maybe the SCADA wasn't integrating correctly. Maybe the sensors were placed wrong. But over four years of reviewing deliverables—roughly 200 unique condition monitoring system proposals per year—I realized the issue was deeper.

The monitoring system itself was often set up to tell you what's easy to measure, not what's actually wrong.

The Deeper Issue: Granularity

To be fair, most solar condition monitoring systems do a decent job at the string level. You can see if a whole row isn't producing. But that's like having a building management system that tells you a fire is in the building, not which room. The gap in granularity is where the real problems live.

I've seen a site where daily string-level data looked perfectly normal—within 5% of expected production for three months. The issue? One of the inverters had a degraded MPPT channel that was micro-cycling. The system-level reports calculated a 3.2% loss, which got flagged as 'within expected variation.'

It wasn't until someone ran a per-module sweep using DC-optimized telemetry that we found six panels underperforming on that string. That minor issue had been accumulating for a full quarter. Extrapolate that across 500 units in a commercial installation, and you're looking at a significant revenue gap.

"I went back and forth between string-level monitoring and module-level for about three weeks on that project. The $18,000 cost difference felt real. Ultimately, the scalability of the data—being able to pinpoint a single panel issue from a single dashboard—is what sold me."
—From a review meeting log, Q1 2023

The Hidden Cost of Missing This

The obvious cost is lost production. But the less obvious one is trust erosion. If your monitoring system keeps reporting 'all good' while small issues compound, eventually a client notices the PPA revenue is off by 5% and starts questioning everything.

I've seen this play out in a project we audited for a commercial client with a 1.2 MW ground-mount system. They had string-level monitoring from a major brand. The production reports showed 95% of expected output—acceptable by industry standards. The client accepted it for three years.

Then they installed a DC-optimized system during a panel replacement. The new monitoring system—component-level—showed that what they thought was a 5% loss was actually 11% in aggregate over that period. The client was furious. The O&M contract had a clause about guaranteed performance, and the discrepancy cost the EPC provider close to $22,000 in credits and rework.

My experience with this case reinforced a principle I now include in every specification review: if you cannot identify a single underperforming panel from your monitoring dashboard, you are not monitoring—you are guessing.

The Solution Is Simpler Than You Think

I don't want to make this sound like a sales pitch, because frankly, the industry is at a point where the technology is mature enough that 'module-level monitoring' is a baseline expectation for any serious commercial installation, not a premium upgrade.

What matters is how the data is structured. A condition monitoring system that uses DC-optimized inverters—like what SolarEdge has been shipping at scale (12.6 GW in 2023 alone)—offers an inherently more granular data hierarchy. You have inverter-level data, optimizer-level data, and panel-level telemetry. If a single panel has a microcrack or a hotspot, you see it in the dashboard within 24 hours.

That granularity isn't just for troubleshooting; it's for verification. If you're working with a commercial client who wants to validate a performance guarantee, having per-panel data means you can prove exactly which module has a problem and what the performance impact is. No arguments. No ambiguity.

The standard you expect from your monitoring system sets the bar for how quickly you can respond when something goes wrong. The issue isn't that string-level monitoring is useless—it's that it's too late. By the time a string-level drop shows up, you've already lost weeks or months of production.

"Per USPS Business Mail 101, the maximum thickness for a standard letter is 0.25 inches. A comparison to a 'flat' envelope (up to 0.75 inches) might seem irrelevant, but the principle holds: if you can't segment the data at the level where the problem occurs, your 'envelope' is holding everything and you lose the ability to pinpoint. Source: USPS Business Mail 101."

Granted, implementing this level of monitoring requires upfront coordination. The installer needs to plan for the data architecture, verify that the telemetry path is robust, and train the O&M team on how to interpret the dashboard. But on a 50,000-unit annual order, the cost of that coordination is a fraction of what one avoidable performance claim costs.

A Personal Note on Decision Fatigue

Even after choosing to specify DC-optimized monitoring for that large commercial project in 2022, I kept second-guessing. What if the price premium was eating into the client's IRR? What if the training overhead delayed the project close? The two weeks between the decision and the first data feed being live were stressful.

But the moment we opened the dashboard for the first time and saw per-panel data flowing—with each module's performance mapped against its expected curve—I relaxed. The client called me the next week and said, 'This is the first time I've actually trusted a production report.' That single sentence was worth the entire effort.

In my experience—which is based on about 200 mid-range to large-scale commercial projects—the decision to invest in granular monitoring is one of the few that consistently pays for itself. The question isn't whether you can afford the data granularity. It's whether your brand can afford the next performance dispute.

If you're specifying a condition monitoring system for a new solar installation, ask yourself: if one panel fails tomorrow, will your dashboard tell you by the end of the week, or will it tell you by the end of the quarter?