PROJECT INSIGHT

The 96% Efficiency Promise That Almost Went Out in a BESS OEM Spec — and How Production-Line Data Caught It

Tuesday morning, 9:14. I opened a BESS OEM spec sheet from a distributor we'd been working with for almost a quarter, and stopped at one line in the performance table:

DC round-trip efficiency: ≥ 96%.

Not because 96% looked impossible. Because there was nothing around it. No temperature. No state-of-charge window. No C-rate. No note about whether that number was measured at the DC bus, on the AC side of the inverter, or with auxiliary loads running.

At Gotion, I approve customer-facing specifications before they go out—especially BESS OEM and private-label work, where a distributor puts their own brand on our cells, packs, and energy storage systems. In a normal year, I review roughly 180 of these documents. Four years of that has taught me one thing: a spec sheet can look precise and still leave the most important sentence unspoken.

This one was a classic.

The spec that looked perfectly fine

The deal itself was reasonable. Five containerized LFP units, roughly 25 MWh in total, under the distributor's own brand for a solar-plus-storage portfolio. Gotion would supply the cells from our own lithium battery production line, build the packs, and handle the hardware manufacturing side. The commercial terms were almost settled. What remained was the performance document—the one that ends up attached to purchase orders, and eventually to warranty claims.

Their engineer had clearly done this before. The document included system voltage architecture, cooling requirements, protection setpoints, communication protocol details, even a storage and handling section. It gave me confidence, right up until the efficiency table.

“DC round-trip efficiency: ≥ 96%.”

I don't think anyone was trying to mislead us. That phrase shows up across the industry, and a lot of engineers copy it from one project to the next. But a number without a boundary is not a specification. It's a hope.

The number had no ruler

Rather than guess, I asked their engineer a simple question: when this system is installed, how would they verify that 96% promise? There was a pause. Then the answer: “We'd measure round-trip energy at the site meter.”

That answer tells you everything. A site meter sits after the transformer, after the inverter, after the cables, and after the thermal management system has been chewing power to keep the container cool. The battery is only one layer in that stack. The same hardware can deliver a very different number depending on where you draw the measurement boundary.

To be fair, this isn't unique to BESS. Every energy storage supplier in the market has their own way of stating efficiency, and most of them are internally consistent. The problem is when a buyer takes a number from one BESS supplier's datasheet and drops it into an OEM contract without the test conditions. That's not a vendor problem. That's a definition problem.

Our quality team sent the spec back with one question written across the top: measured from where, to where, at what temperature, at what state of charge, and with which loads included?

What the production line told us

Here is where having our own cell manufacturing changes the conversation. We didn't have to argue about what was theoretically possible. We pulled actual data from Gotion's lithium battery production line and pack integration records, then compared their spec against measured performance under different boundary conditions.

The result wasn't dramatic. There was no smoking gun, no obvious deception. That's what made it dangerous.

Under one set of conditions, the system could approach the number their spec promised. Under another set—the one their commissioning contractor would actually use—the number would fall short. Both measurements could be described as “DC round-trip efficiency.” The phrase was the same. The ruler was different.

The call ended with a sentence I've heard many times since: “I assumed every serious supplier tested the same way.” That assumption is exactly what causes these problems. Learned never to assume that after this one.

What we changed before signing

We didn't tell them they couldn't have 96%. We told them they could have exactly that, but only if we agreed on what it meant. The fix was a redline document—three extra pages of measurement and test procedures that now sit inside every BESS OEM proposal we send.

  • Measurement boundary: which hardware is inside the efficiency number and which is outside.
  • Ambient temperature: whether the test happens at 25°C or over a range that reflects real operating conditions.
  • State-of-charge window: whether the test starts at full charge and ends at empty, or stays within the BMS's recommended operating band.
  • Charge and discharge rates: a number at 0.25C is not the same as the same system at 0.5C or 1C.
  • Auxiliary loads: HVAC, controls, and station power included, or intentionally excluded.
  • Meter class: because a 0.5% meter and a 2% meter can disagree enough to decide a commissioning test.

Their engineer initially pushed back. Then we walked through the consequences of leaving the spec vague. If the system had shipped and the energy yield came in below the modeled value, the first fight would have been about which one of us was “wrong.” The second fight would have been about money. The third, about schedule.

Dodged a bullet, honestly. The distributor later told us their commissioning contractor had already planned to verify performance from the revenue meter. That would have failed almost any large-scale BESS on a cold morning or a hot afternoon, no matter whose name was on the container.

We didn't have a formal verification protocol for every private-label BESS spec at the time. We relied on individual reviewers having enough experience to ask the right questions. That worked until it almost didn't. After this order, I built a checklist for the team. Should have done it after the first near-miss.

A short lithium battery specification guide for anyone buying BESS OEM

If you're a distributor, an integrator, or a project developer evaluating a BESS supplier, you don't need to become a cell engineer. You need to ask for the conditions behind each claimed number.

The most common pitfall is capacity. A cell's rated capacity depends on discharge rate, temperature, end-of-discharge voltage, and whether you're talking about beginning of life or end of warranty. The same applies to system energy: rated energy is not necessarily usable energy, and usable energy depends on the state-of-charge window the BMS actually allows.

The second pitfall is efficiency. Before you compare two suppliers, force both of them to state the measurement boundary in writing. If one says AC round-trip and the other says DC round-trip, you're comparing apples to oranges. If auxiliary loads are missing from both, you're comparing two optimistic hypotheticals.

The third is cycle life. A lithium battery specification guide will often list a cycle number, but that number means nothing without a depth of discharge, a temperature, a C-rate, and an end-of-life threshold. A cell that delivers 8,000 cycles at 25°C and 50% depth of discharge will not deliver the same number in a hot container at 90% depth of discharge.

The pattern is always the same. The numbers aren't necessarily false. They're just incomplete.

Prevention is cheaper than correction

What struck me most about this order was how cheap the fix was. Three days of back-and-forth, a few pages of definitions, one phone call. The cost was almost nothing. The cost of getting it wrong would have been a failed acceptance test, a delayed project, a damaged warranty relationship, and a distributor who learned the hard way why “96%” on a datasheet is not a performance guarantee.

5 minutes of verification beats 5 days of correction. In this case, it was more like 5 weeks.

The systems shipped last month. I checked the final spec before it went to the printer—the distributor's logo on the cover, our test conditions in the appendix, and a number that finally meant the same thing to everyone involved.

If you're shopping for a BESS OEM partner, don't just ask what their system can do. Ask how they measure it, and what conditions the guarantee is written against. If their engineering team pushes back on ambiguity, that's a good sign. If they sign a vague spec without asking a single question, be suspicious.

The lithium battery production line data doesn't lie. But you have to know which ruler it's using.