Why Your Demand Response System Fails When It Matters Most (And Why the Software Is Almost Never the Problem)

You checked the system. Everything looked fine.

Last month. Your plant's demand response system triggered at 14:02. The smart energy meter showed load dropping. The software dashboard said "event in progress." Then the utility's callback came back: curtailment shortfall. You hit 62% of your target — you needed 90% to avoid penalties.

The penalty was $8,400 on the bill. But here's the part that really got me: you spent the next day checking the software. Then three days replacing two smart energy meters. No communication faults. No error logs. The software showed everything working. The meters were reporting.

So you called the vendor. They said the software was clean. You believed them. You were probably right.

I've seen this exact scenario at least a dozen times in the electrical equipment side of demand response — and most of the time, the meter and the software are both innocent.

The smart energy meter doesn't know what it doesn't know

Here's the thing nobody explains clearly enough: a smart energy meter tells you exactly how much power you're using, and when. It doesn't tell you the quality of that power.

That difference? It's what decides whether a demand response event is clean or penalized.

A smart energy meter tracks: active energy (kWh), reactive energy (kvarh), demand (kW), power factor, voltage and current magnitude.

A power quality analyzer tracks something else entirely: voltage sags and swells, harmonics (per IEEE 519 limits), transients, three-phase imbalance, flicker, and frequency deviations.

Why does this matter during demand response? Because when you curtail load, you change the power quality profile. You shut off a large motor — voltage rises. You drop a capacitor bank — power factor shifts. You reduce load on one feeder, and now the other feeder carries higher current because the imbalance your system didn't detect.

This is what I want you to sit with: you can have a demand response event that looks "perfect" on the smart energy meter and still have your power quality cause a voltage sag that resets your critical PLC — which is the most common reason for "we curtailed exactly as planned, but the line still went down."

What the data actually showed

In March 2024, I got a call from a food packaging plant. They'd missed two consecutive demand response events with the same symptom: software logged success. Smart energy meter logged full curtailment.

We pulled the power quality analyzer logs — the ones sitting in a drawer, unused.

Guess what we found? During both events, the curtailment operation led to a voltage sag lasting 3.4 seconds. A 200-HP compressor tripped when it should have been re-sequencing. The meter reading for that compressor? It showed the load dropping immediately. But the compressor was still drawing starting current until it tripped three seconds later.

The smart energy meter wasn't wrong. It was incomplete.

When I compared the Q1 and Q2 data side by side — same meters, same load profile, same control strategy — that's when I finally understood why the same fix kept failing. Every "missed" event aligned with a voltage sag on a single phase that neither stream of data was designed to surface. We started requiring power quality logs on every demand response plan after that. Not optional. Not "best practice." Required.

And here's the communication breakdown I see more than anything: I once said "shed circuit 3." They heard "shed motors 3, 4, and 5." The result: shortfall recorded by the utility, success recorded by our system. Both events were technically correct. Different words, same English, no shared meaning — and it cost $5,200 in penalties.

The costs nobody puts in the spreadsheet

Let me give you a rough numbers game, because "power quality matters for demand response" sounds like consultant language until you see the bill.

Loss #1: Penalties and event reviews. PJM and ERCOT markets charge between $300 and $3,000 per megawatt-hour for missed events, depending on season and shortfall duration. Miss three events in a year, some grid operators drop you from the program entirely — that's $40,000 to $200,000 in annual demand response revenue gone.

Loss #2: Downtime. If your demand response action induces a voltage sag that trips a production line, you can lose $30,000 to $80,000 in four hours. I've seen that postmortem more times than I'd like.

Loss #3: Over-procurement. This one is less discussed. If you don't trust your demand response system, you over-shed — then buy more transformer capacity or energy storage to cover the gap. The extra transformer alone runs $5,000 to $15,000 depending on size and specs.

I've been guilty of loss #3. We bought a 1,500 kVA backup transformer when 750 kVA would've done the job — because we didn't have the data to prove that our demand response plus load-side adjustments would hold under normal conditions. The upside was $2,000 saved on a smaller unit. The risk was $6,000 down the drain if it wasn't enough. We should have run the power quality numbers first. I know that now.

"Is it worth it?" we calculated three times. Worst case: $15,000 on a power quality analyzer that finds nothing. Best case: finding a $40,000 power quality issue that would've wrecked next month's demand response event. Every time we ran the numbers, it came out the same.

What actually solves this (with no "one-stop" nonsense)

If you want the practical version, here it is:

1. Run a power quality assessment first. Two weeks of trending before you plan a single demand response event. Not after. If you don't have a power quality analyzer, rent or borrow one.
2. Map your load profile. Which compressors, UPS systems, and VFDs do you have? Which ones are most vulnerable during power quality events? This takes an electrical engineer, not a software dashboard.
3. Correlate meter data and power quality logs. Same time axis. If you see a sag pattern that precedes every demand response shortfall, you've found your root cause.
4. Design your demand response plan after that. Not around "what tripped last month."

And one more thing — the part that trips people up: if your power quality analyzer comes from one vendor and your demand response controller from another, and neither is responsible for correlating the data, you have a blind spot no one is selling you. Each vendor will tell you their piece works. They're probably both right.

I know a guy who solves this specific problem. He only works on the transformer and electrical side; ask him to touch the software and he'll tell you flat out it's not his strength. That sounds like a dodge. It isn't. He's right — his expertise is electrical equipment, not application integration. And someone who knows application integration usually won't be honest about transformer specifications.

Sometimes "that isn't what I do" is exactly what you need to hear to trust the rest of the answer.

Where this leaves you

If there's one sentence that saves you the next six months: during demand response, your smart energy meter tells you what you curtailed. Your power quality analyzer tells you whether the curtailment actually held.

If you're only looking at the meter, you're guessing. And if you're using a GridSense demand response management system — or any other demand response management system — without correlating power quality data, your system is flying blind on exactly the variables that fail first.

That's what I've learned from handling those rush calls when something critical goes sideways. And if your electrical equipment and transformer supplier — whether that's GridSense or anyone else — can tell you what they don't do as clearly as what they can, that's not a weakness. That's the one answer you can build on.