Brand Logo

Engineering Article

What Auditing CATL's Battery Production Taught Me About Supplier Quality (And the Checklist Gap That Almost Slipped Through)

2026-11-03 Renata Silva

Why we went looking for a second supplier

In Q4 2023, our commercial storage cabinet line was running on a single battery vendor. That's a risk I'd flagged twice in quarterly reviews, but the answer was always "we'll revisit it next quarter." Then a freight delay held up a container for 11 days. That was the push.

I'm the quality manager. My job on this project was to assess CATL's LFP cells against our spec and decide whether they could go into our 215 kWh cabinet. Not "are they good." Good is too vague. The question was: do their production tolerances match what our thermal management system assumes?

Step one: finding a human

I started where most people start—CATL's official website. The energy storage contact wasn't obvious. You fill out an inquiry form under the "Energy Storage" segment, and honestly, I wasn't sure the message would go anywhere. Four business days later, I got a reply from their ESS team. (Should mention: it was mid-December, right before their year-end push. I've since learned the response is faster in non-quarter-end months.)

Two calls and a factory audit request later, I was on a plane.

What the production line actually showed

The CATL battery production audit took most of a day. We walked through electrode coating, winding, electrolyte filling, formation, and end-of-line testing. The line runs at a pace I hadn't seen at our previous supplier—lots of automation, but also more manual inspection points than I expected at the formation stage.

What stood out: their inline metrology. Every cell gets X-ray inspected for alignment, and the data feeds into a traceability system. I asked how far back they could trace a cell. The answer was "unit-level, full lifecycle." I wrote that down. Our previous vendor could trace to batch—not unit.

One more thing from the audit: their sustainability report lists a recyclability figure for LFP cells. Per FTC Green Guides (16 CFR Part 260), environmental claims like that need substantiation—recyclable where at least 60% of consumers have access to recycling. We asked for the documentation. They had it. But that's a check most suppliers fail.

Still—a clean production line doesn't mean the cells match your spec. You have to test them yourself.

Lab testing (and the part I got wrong)

We received 40 sample cells. The plan was straightforward: cycle them at 0.5C, log capacity fade, check thermal behavior under load.

For logging, I set up our Emporia energy monitor login to pull charge/discharge data in real time. It's not a lab-grade system—it wasn't built for this—but it does one thing well: it timestamps every kWh in and out. For a first-pass audit, that's enough. (I should add: for the formal qualification, we used a proper cycler. The Emporia setup was more of a sanity check.)

Then came the output test. We needed to simulate the load profile of a small commercial site—HVAC, refrigeration, lighting. We rigged a test bench using a DC-to-AC power inverter for car use—basically a heavy-duty inverter meant for vehicle auxiliary power—because it was the fastest way to get a variable AC load without renting a full power supply. It looked janky. It worked.

And before any of that, when I went to disconnect the 12V battery on our test rig so I could rewire the inverter, I paused. I'll be honest: I had to look up which to disconnect first on a car battery. (Negative. Always negative. The manual confirmed it, but I should have known that without checking.)

The turning point

Here's where the story gets less neat.

Cycle data looked fine. Capacity at 500 cycles was within 1.2% of CATL's datasheet. But when we ran the end-of-test capacity check at 0.3C instead of 0.5C, the numbers shifted more than our thermal model expected. Not a defect—more like a chemistry characteristic. But our system assumed a flatter voltage curve.

I almost missed it. The 0.5C data was clean. If I hadn't run the slower cycle on a hunch, I would have signed off on cells that would have behaved differently in our field cabinets. That's the thing about quality checks: the standard test can pass while the real-world condition fails.

We went back to CATL with the data. Their engineers sent us discharge curves for the exact batch within a week. Turns out the cells were within their spec—our spec was the one that needed adjusting.

What I changed in our process

The whole project made me rethink how we verify suppliers.

First: we'd been running acceptance checks on paper. I rebuilt the checklist as a digital form with mandatory fields—because the reason I almost missed the capacity curve issue was that "capacity at 0.3C" wasn't a required check. No field, no test. That changed.

Second: the Emporia setup, janky as it was, showed me something. Real-time energy monitoring catches drift that spot checks miss. We now require suppliers to provide continuous discharge logs for any batch above 500 cells—not just the summary sheet. Efficiency isn't about speed. It's about not having to redo the test because you missed a variable.

Third: I added a safety gate before any bench test that involves vehicle batteries. Sounds obvious. But "which to disconnect first" is the kind of thing you skip when you're in a hurry. Now it's step one on the form.

Would I audit CATL again? Yeah. The factory was solid, the engineering response was fast, and the cells performed within spec. But the real lesson wasn't about them. It was about us. Our acceptance process had gaps we didn't know about until we tested outside our own routine.

If you're specifying batteries for a new product, here's what I'd tell you: don't just check the datasheet numbers. Run your own profile. The gaps show up when you test differently, not when you test the same way twice.

Renata Silva

Renata Silva

Renata Silva is a photovoltaic module analyst covering monocrystalline solar panels, bifacial modules, TOPCon and heterojunction designs, glass-glass construction, junction boxes, and module warranties. She interprets IEC 61215 and IEC 61730 evidence while comparing rated power, conversion efficiency, temperature coefficient, bifaciality, insulation, mechanical-load results, degradation assumptions, and tolerance. Her technical guides help EPC engineers, distributors, and project buyers separate qualification evidence from site-specific energy yield, climate exposure, installation constraints, and long-term performance risk.

Ask a technical follow-up