Solar + BESS Round-Trip Efficiency: Where the kWh Go

A realistic teardown of where a solar-charged battery's energy goes between the PV module and the grid. If you are modeling a utility-scale or behind-the-meter hybrid, knowing the component efficiencies -- and which ones are optimistic in a vendor spec sheet -- is the difference between a bankable IRR and a re-forecast at year two.

The energy chain

PV module -> DC bus -> DC/DC converter -> battery -> DC/DC converter -> DC/AC inverter -> MV transformer -> collection cable -> POI
(or for AC-coupled: each side of the battery has its own DC/AC inverter)

Topology choice: DC-coupled vs AC-coupled

Feature	DC-coupled	AC-coupled
Typical round-trip eff.	84-89%	78-84%
Clipping recapture	Yes (charges from clipped DC)	No (clipping already lost at inverter)
ITC treatment (U.S.)	Both qualify; DC-coupled sometimes contested pre-IRA; now clear under IRA for hybrid facilities	Both qualify
Retrofit to existing PV plant	Harder (new PCS shared)	Easier (drop in parallel)
Capex	Usually lower for greenfield (one PCS)	Higher (two PCS)
Independent dispatch	Solar and battery share one inverter; interaction	Fully independent

Rule of thumb: DC-coupled if the PV DC/AC ratio is > 1.3 and clipping > 3% (grab the clipped energy). AC-coupled for retrofits or when independent dispatch matters more than a few % of clipped kWh.

Round-trip efficiency bookkeeping (DC-coupled example)

Starting with 100 kWh of DC energy from the PV array:

Step	Component loss	Typical efficiency	Energy remaining
PV to DC bus	Wiring, combiner, dust	98%	98.0
DC/DC to battery	Charge converter	98%	96.0
Battery charge (LFP)	Coulombic + IR	96%	92.2
Idle self-discharge (1 day)	Leakage + BMS aux	99.5%	91.7
Battery discharge	IR, temperature	96%	88.0
DC/DC from battery	Discharge converter	98%	86.3
DC/AC inverter	PCS, CEC weighted	98%	84.6
Aux loads (HVAC, BMS, controls)	2-4% of throughput	97%	82.1
MV transformer + cable	Copper + core	99%	81.3

End result: ~81% round-trip from PV to POI. Subtract a further 1-2% for capacity fade in year 5, and your spreadsheet should use 78-80% when pricing long-duration arbitrage.

Where vendor specs lie (or at least, mislead)

"95% RTE" on a battery pack. That's usually the DC-to-DC round-trip at 25°C, C/3 rate, beginning of life, excluding BMS auxiliary loads. Real-world AC-to-AC is 10+ points lower.
Inverter efficiency at peak load. CEC-weighted efficiency is what actually matters in simulation; peak efficiency numbers can be 1-2% higher than the CEC value.
Thermal system losses. Liquid-cooled batteries draw 2-5% of rated power continuously for the chiller when operating in hot climates. This kills storage economics in Phoenix or Riyadh vs. Germany.
Self-discharge and parasitic load on the diesel backup/auxiliary. If your BESS site has a diesel backup running weekly for tests, add its fuel in the opex.

Capacity fade -- the silent IRR killer

LFP cells typically guarantee 70% of beginning-of-life capacity at 6,000 cycles (or 15-20 years) at 25°C, C/2 cycling, 0-90% SOC window. Real conditions rarely match:

Each 10°C above 25°C roughly halves calendar life (Arrhenius).
Using a wider SOC window (0-100%) accelerates fade by ~30%.
Frequent 1C+ charging for frequency response drops the 70% point to ~4,000 cycles.

Design in augmentation capacity from day 1 -- either by starting oversized (20-30% extra MWh) or by leaving floor space and MV slots to bolt on new cabinets at year 5 and year 10.

Design heuristics for modelers

Round-trip AC-to-AC year 1: 85% for DC-coupled LFP, 82% for AC-coupled LFP, 80% for AC-coupled NMC.
Year-10 fade factor: 0.80 unless augmenting.
Auxiliary load draw: 3% of rated power during operation in temperate climate, 5% in hot climate.
Inverter clipping recovery for DC-coupled BESS (DC/AC ratio > 1.3): +2 to +4% annual energy vs. a PV-only system.
Degradation warranty: get it in writing, tied to usage envelope.

Quick math check

A 100 MW / 400 MWh LFP BESS in Phoenix, DC-coupled to a 150 MWdc PV plant.
Year 1 throughput (1 cycle/day): 400 MWh charge -> 340 MWh delivered AC (85% RTE).
Year 10 (6.2% annual fade, no augmentation): 400 MWh charge -> 270 MWh delivered AC.
Aux load in Phoenix (5% x 100 MW x 4 hr): 20 MWh/day parasitic.
Effective RTE year 10, all-in: ~62%.
Build the pro-forma with this number, not the 95% on the cover page.

References

NREL Annual Technology Baseline (2025) -- Utility-scale battery storage cost and performance.
Sandia / DOE "Energy Storage Performance Testing Guidelines" 2020.
IEC 62933-2-1 -- Electrical Energy Storage Systems: Unit Parameters and Testing Methods.
IEEE 2030.2.1 -- Guide for Interoperability of Energy Storage Systems Integrated with the Grid.
EPRI "Battery Energy Storage System Performance Modeling" 2024.
Lawrence Berkeley National Lab (LBNL) "Capacity Fade in Lithium-Ion Batteries" series.