March 2024 — the batch that shouldn't have passed
In March 2024, I was running final quality review on a photovoltaic module batch headed for our Uzbekistan project. The supplier was one we hadn't worked with much before.
Their documentation looked fine. IEC 61215 and 61730 certificates in order. Factory audit scores above our threshold. Warranty terms matched what our project leads needed. The sample IV curves they sent over were clean — fill factor above 79%, temperature coefficient where we expected it.
On paper, nothing was wrong.
Then we ran EL imaging. Microcracks across the backsheet. Not catastrophic, but visible. The supplier's response came back fast: "within IEC standard tolerance."
They were right, technically. IEC 61215 allows a certain level of microcracking if it doesn't lead to measurable power degradation. But "compliant with IEC" and "ready for 25 years in Central Asian temperature swings" are two different thresholds. Our site ranged from -25°C to +45°C. Any microcrack that shows up in factory testing becomes a stress point.
We rejected the 12 MW. The supplier redid it at their cost — that was the good news. The bad news is we almost didn't catch it. Our original checklist only verified certificates and spot-checked IV curves.
What I initially got wrong about "Tier 1"
When I first joined the quality compliance side at ACWA Power, I assumed Tier 1 BNEF status meant everything. Tier 1 equals best. Simple.
I held that assumption for about six months.
Tier 1 is a bankability measure used by financial institutions. It reflects a manufacturer's balance sheet, project pipeline, and financing track record. It is not a commitment about PID performance, LID, or field degradation.
I've watched Tier 1 modules from one manufacturer underperform Tier 2 modules from another in damp-heat chamber testing in our Dubai lab. That's not a contradiction. The ratings measure different things, and I was treating a finance metric like a quality score.
These days we look at BNEF ranking for context, but we don't treat it as proof of quality. What actually predicts performance is the PVEL scorecard, where you can see pass/fail results by test category — damp heat, mechanical load, PID, LID+LETID.
The scoring card we built after that batch
After the 12 MW incident, the team and I rebuilt the evaluation approach. Three different tables:
- Documentation review — IEC 61215, IEC 61730, ISO 9001/14001, factory audit reports, and PVEL scorecard results where available
- Physical sample testing — EL imaging, IV curve sweep, and damp-heat chambers at an independent lab on every sample batch
- Total cost model — not just the per-watt price, but shipping, duties, warranty cost, replacement logistics, and the most underpriced line item of all: the cost of bad modules arriving on site
The third table changed the most, and I should have built it years earlier.
Rejecting that 12 MW batch didn't cost us anything directly — the supplier absorbed it. But it added 19 days. That 19 days pushed the project energization into a quarter with less favorable financing, and re-shuffled our whole model.
Financially, we still made the right call. But the per-watt price alone would have told a different story. It would have pushed us toward the cheaper supplier and then buried us in the back end.
What the hidden costs taught us
When I compare vendor quotes now, I break TCO into four buckets:
- Module price — the per-watt number that everyone thinks is the whole picture
- Logistics — factory to site, including insurance, customs, and inland transport
- Warranty execution — not the warranty number, but what it actually costs to enforce if 3% of modules fail in year five
- Risk premium — how much buffer you need to hold for a supplier with no reliable track record
A supplier quoting $0.02/W cheaper can end up $0.05–0.08/W more expensive once you factor in re-work, field labor, or grid-connection penalties. That's not hypothetical. I remember one project with a no-name module that clocked roughly $400,000 over twelve months in replacements and associated construction labor.
We didn't own that project, but we were the ones helping clean it up.
What we actually look at in a photovoltaic module supplier
If someone asks me what to look for, I give four items and they're in order:
- Actual project performance in the last 24 months. Ask for references with site names. Not general assurances.
- Factory audit consistency. Not a triennial certification. Are there surprise visits? Who runs them?
- Warranty enforceability. Who actually backs it, where, and can it survive 15 years from now?
- Capacity reality. Can they actually deliver 200 MW on contract, or will part of it get subcontracted to a facility you don't know about?
The solar module private label angle deserves specific caution here. With private label deals, the brand is the one that carries the quality reputation, but the brand has little day-to-day control. Don't white-label a factory you can't audit yourself, unless the contract has serious teeth.
What I'd do differently if I started over
Honestly, roughly what we do now — just earlier:
Build the scoring card before anyone sends you samples. Test each sample batch through third-party EL, damp-heat, and PID protocols. Model the delay cost at quote stage, not at execution stage.
Most of all: I used to think of the lowest quote as the starting point and adjust up from there. Now I think of it as the last input into a broader table of continuity and reliability.
In renewable projects, the tariff has gotten thin enough that $0.02/W looks enormous. What actually moves the needle is when the $400,000 O&M surprise shows up and the $0.02/W was irrelevant — because the modules worked.
What I'd tell my earlier self
Certificates prove minimum compliance, not quality. Tier 1 is a finance rating, not a performance guarantee. The cheapest quote is very rarely the cheapest project.
After the 12 MW batch, we changed our supplier review process. We haven't rejected a shipment since — not because suppliers got better, but because we finally know what we're testing for.
If you're evaluating photovoltaic module suppliers right now, my suggestion is the same you've heard with a different weight behind it: write the scoring card down. Convert assumptions into tests. Then look at the real cost, not the quote.
It won't eliminate risk. It will keep you from repeating the mistakes I made.
