Round-trip efficiency RTE

What it is, precisely

RTE equals energy out divided by energy in over a defined cycle, with numerator and denominator measured at the same boundary and under stated conditions, or the number means nothing.

Three boundaries recur in grid-scale work: the DC bus at the battery rack, capturing only cell coulombic and resistive losses; the AC terminals of the PCS, adding inverter conversion loss; and the POI or revenue meter, adding medium-voltage transformer losses plus station auxiliary and parasitic consumption. Each step down the chain strips roughly one to three points off the headline figure, so DC-side RTE always reads higher than net AC.

Round-trip and one-way efficiency are related but not interchangeable. To a first approximation RTE is the one-way efficiency squared, because energy runs through the same loss chain twice: about 95 percent each way lands near 90 percent round trip. This is why a PCS advertised at 97 percent still burns roughly six percent of throughput over a full cycle on its own, before cell, transformer, and HVAC losses are counted. Keeping the AC-DC-AC path and the two-pass math straight is the difference between a defensible number and a misread datasheet.

Because RTE moves with C-rate, depth of discharge, temperature, and dwell time between charge and discharge, a credible figure always travels with its test conditions. A 4-hour system cycling at 0.25C sees far lower resistive losses than the same hardware pushed at 0.5C, since loss scales with current. A net AC RTE at the POI, at a stated C-rate and ambient temperature, with HVAC running, is an engineering number you can bank; a bare percentage floating on a brochure is a marketing one you cannot.

Why it matters in a real grid-scale project

RTE is a direct revenue and cost driver, so it belongs in the financial model, not just the spec sheet. Every kWh lost to inefficiency is energy you paid to charge but can never sell, so RTE flows straight into arbitrage margin: the effective charging price is the market price divided by RTE. The arithmetic is unforgiving, as delivering 100 MWh at 85 percent RTE means buying about 118 MWh. For a project cycling daily, a one-to-two point swing compounds over thousands of cycles and shows up plainly beside Availability and Degradation as the parameters lenders scrutinize hardest.

It also governs the parasitic and thermal design. The gap between DC and AC RTE is dominated by PCS conversion loss and station auxiliary load, above all the HVAC that keeps LFP cells in their optimal temperature window.

Hot climates push auxiliary consumption up and net RTE down, which is why HVAC sizing, transformer selection, and the self-consumption strategy are RTE decisions, not merely thermal ones. Liquid-cooled containers became standard partly because tighter cell temperature control lifts both efficiency and cycle life, tying RTE to Cycle aging and Calendar aging through the same thermal envelope.

Typical values and standards

For modern LFP utility-scale systems, DC RTE is commonly quoted near 92 to 96 percent, AC RTE at the PCS terminals near 88 to 93 percent, and net AC RTE at the POI near 85 to 90 percent once transformer and auxiliary loads are included.

NMC systems sit in a similar band; the differentiators are PCS efficiency, auxiliary load, and operating temperature more than chemistry. For contrast, vanadium flow runs about 65 to 75 percent AC and pumped hydro about 70 to 80 percent, a core reason lithium-ion dominates short-duration economics. Treat every figure here as an accepted range, never a guarantee.

Commissioning RTE tests generally reference IEEE 2030.3 and IEC 61427-2, which fix how charge and discharge energy are measured and at which boundary so results are repeatable and comparable to the warranty; IEEE 2030.2.1 frames how the BESS is designed and operated around those numbers.

RTE is a performance metric, distinct from the safety stack an engineer also lives with: UL 9540 (the ESS product safety certification), UL 9540A (the fire-propagation test method that feeds NFPA 855), and NFPA 68/69 (deflagration protection). Never conflate UL 9540 with UL 9540A; a good commissioning campaign verifies performance and safety in parallel.

RTE is not constant over project life. As cells age, internal resistance grows, resistive losses rise, and RTE drifts down, typically a point or two from beginning of life to End of Life, in parallel with Capacity fade. That drift is separate from usable-energy Degradation: RTE tracks how efficiently energy passes through, while Capacity fade tracks how much can be stored at all. Both feed the Augmentation plan, because a fleet losing capacity and efficiency together needs more charging energy per delivered MWh every single year of the contract.

How it shows up in specs, studies and contracts

On a datasheet, ask the same five questions every time: at which boundary is the number quoted, at what C-rate, at what ambient temperature, at what depth of discharge, and are auxiliaries included or metered separately? Vendors quote the flattering boundary by default, so a container sheet showing 95 percent almost certainly means DC-side without HVAC.

Demand the full loss budget: cell losses, the PCS efficiency curve at your actual dispatch point, transformer no-load and load losses, and an annual auxiliary energy estimate for your climate. PCS efficiency drops steeply at low load, so a unit dispatched at 30 percent of rating never reaches its peak number.

In contracts, RTE appears as a guaranteed performance parameter beside the Capacity warranty and the Availability guarantee. EPC and equipment supply agreements name a tested net RTE, a test protocol, and liquidated damages if the measured value falls short, so pin down the measurement boundary and any temperature-correction curve before signing.

Expect a commissioning capacity-and-RTE test and often periodic retests, with a tolerance around one percentage point as common negotiating ground. In merchant markets the loss is settled implicitly instead: charging energy is bought at the revenue meter, and every efficiency point is simply borne by the owner.

Common pitfalls

The classic error is comparing RTE figures quoted on different bases: a 95 percent DC number and an 87 percent net AC number can describe the very same hardware, not a better product. Never mix DC-DC and AC-AC figures in a financial model without an explicit conversion, and never assume auxiliary power is included, because some suppliers meter station service separately and exclude it from the guarantee entirely. When a bid looks two points better than a rival, check the boundary before you credit the technology.

Operational RTE also diverges from test RTE. A frequency-regulation duty cycle with shallow cycling, partial-load PCS operation, and long idle periods carrying HVAC load can post a fleet-average RTE well below the commissioning result, because the test cycle is a clean full charge and discharge at favorable conditions. When benchmarking real plant data, define the window carefully: idle-day auxiliary consumption dragged into the calculation can make RTE look catastrophic even when every piece of equipment is performing exactly to spec.