Beginning of Life BOL
Beginning of life is the performance point a grid-scale battery actually starts its commercial life from — the capacity, energy and internal resistance it has when the revenue meter starts counting, not the number printed on the datasheet.
Between cell manufacture and the commercial operation date sit months of transit, storage, installation and commissioning, all of it calendar aging, so BOL energy lands measurably below the factory measurement — and usually below the label — on a plant where nothing has gone wrong at all.
The COD capacity test is what turns BOL from a concept into a number: a witnessed discharge at a defined meter under defined conditions, whose result becomes the baseline every later test and warranty argument references. And because every retention percentage, state-of-health figure and end-of-life floor in the contract stack is a fraction of BOL, the definition of that denominator is one of the most consequential lines in the supply agreement.
Reviewed August 2026 by Sergey Syrvachev
New to BESS? Start free with the 7-email fundamentals course — no cost, no account.
What it is (precise)
BOL names the initial state of the asset — its capacity, energy and resistance at the start of service life — and the first job is deciding which zero that means. There are three candidates. The factory measurement after cell formation is the vendor's fingerprint and the source of the nameplate rating. Delivery to site is where possession, and often risk, changes hands.
And the COD capacity test is where the plant's performance is measured as a plant, at a contractual boundary. In a grid-scale contract the working definition is the last one: BOL is measured performance at a stated meter, under stated temperature, rate and SOC-window conditions, at the start of commercial operation. Nominal energy is a rating; BOL is a test result. The two are related but never equal, and the gap between them is real energy someone has to account for.
The gap has two components. The first is time: the months between the factory measurement and the acceptance test sit on the steepest, approximately square-root-of-time portion of the calendar-fade curve — on the order of 2-4% total in the first year for LFP — so a plant that takes about a year to reach its test arrives a couple of percent below the factory number with every cell behaving exactly as modelled.
That window, who owns it and how contracts allocate it is the subject of Pre-COD degradation. The second is boundary: BOL exists at every link of the energy chain — nameplate DC at the racks, usable energy inside the SOC window, AC energy at the point of interconnection, where usable AC typically lands around 85-92% of DC nameplate at BOL. Quoting BOL without naming the link is how baselines end up in dispute.
The COD capacity test fixes it
Operationally, BOL is established by one event: the capacity test in the commissioning sequence. The protocol is contractual — typically a full charge, a rest period, then discharge at rated power across the contractual SOC window with energy integrated at the revenue meter, temperature-corrected per the contract's rules. The meter itself is a revenue-class instrument — commonly class 0.2S/0.5S under IEC 62053-22 or class 0.2/0.5 under ANSI C12.20 — because at contract scale a fraction of a percent is real money.
Test methodology for storage performance is standardized in the IEC 62933 series, which contracts often reference, but the binding text is the contract's own exhibit: it names the meter, the corrections, the rest periods and the auxiliary-load treatment. COD tests commonly require demonstrating roughly 95-100% of the contracted MWh at the point of interconnection.
The result does double duty. It decides whether the plant passes acceptance — with cure periods, liquidated damages or added racks waiting behind a shortfall — and it becomes the BOL baseline that every subsequent capacity-maintenance test is compared against for the next 15-20 years.
That second role is why the test record matters as much as the test result: the ambient temperature, resting SOC, correction factors and measurement point all need to be documented well enough that the baseline can be reproduced and defended in a dispute a decade later. A BOL number that cannot be reconstructed is a baseline that exists only as long as nobody challenges it.
Between cell manufacture and COD sit months of transit, storage, installation and commissioning, all of it calendar ageing on the steepest part of the curve — on the order of 2–4% in year one for LFP. Acceptance is commonly ~95–100% of the contracted MWh demonstrated at the POI.
- Definition
- Measured initial performance at a stated boundary and conditions — a test result, not the datasheet rating
- How it is fixed
- COD capacity test: full charge, rest, discharge at rated power across the contractual SOC window, integrated at the revenue meter and temperature-corrected
- Typical acceptance bar
- Commonly ~95-100% of the contracted MWh demonstrated at the point of interconnection
- Calendar fade before COD
- Calendar fade between manufacture and COD — on the order of 2-4% in year one for LFP, the steepest part of the curve
- What it anchors
- Retention %, SOH and EOL are fractions of BOL; end of life is commonly 65-70% of BOL usable energy
- Boundary ladder
- BOL exists at every link — nameplate DC, usable energy, AC at the POI (usable AC typically ~85-92% of DC nameplate at BOL)
- Design margin at BOL
- DC overbuild of roughly 10-25% versus the day-one contract quantity, with augmentation sized to the worst guaranteed year
- Not the same as
- Nominal energy (a rating), COD (a legal date), warranty start (delivery, energization or COD — a separate lever)
The denominator of everything after it
Retention percentages, state of health and end-of-life thresholds are all fractions, and BOL is the denominator — "80% retention in year 12" means nothing until the 100% is named. The candidates differ: factory nameplate, the contracted energy, or the measured COD result each put the same physical plant at a different percentage on the same day.
End of Life is commonly set at 65-70% of BOL usable energy, so moving the baseline moves the EOL year without touching a single cell. The direction of the error matters commercially in both directions: a warranty tracked against nameplate when the COD test came in above nameplate starts with margin already banked for the supplier, while a guarantee referenced to raw nameplate at the test date can fail on calendar fade that arrived precisely on schedule.
BOL also anchors the design margin. Projects carry a DC overbuild of roughly 10-25% at BOL versus the contracted energy, sized so the faded fleet still clears the guarantee in the worst contracted year — the distance between the BOL measurement and the contract number is the project's whole fade budget. One separation worth keeping sharp: the BOL baseline and the warranty clock are different levers.
Warranty periods can start at delivery, at energization or at COD, and on a slipping project a delivery-referenced warranty burns years while the BOL test has not yet been run. Setting the capacity baseline at the test date does not by itself decide when the warranty term starts; both need their own line in the agreement.
Common pitfalls
The standing error is reading nameplate as BOL. A financial model that starts the degradation curve at 100% of nameplate on the COD date has quietly handed the plant back the energy it already lost in transit and storage — the curve's year zero was the factory, the model's year zero is COD, and the offset is either absorbed in the vendor's margins or silently missing.
Its twin is boundary mixing: a BOL figure quoted in DC MWh at the racks compared against a guarantee written in AC MWh at the POI differs by the entire loss chain between the two buses, and the comparison flatters whoever chose the bus.
The subtler trap is treating BOL as a synonym for new or for perfect. It is neither — it is a measured state, and it is the best measurement the plant will ever produce, because from that point capacity only ratchets down until augmentation adds hardware.
Vendors' retention curves sometimes label their own factory zero as BOL, so a curve and a contract can use the same three letters for different points months apart; check which zero each document means before putting them in the same spreadsheet. And a strong COD result is not slack: energy above the guarantee at BOL is the headroom that pays for the steep first years of fade, not margin to trade away in negotiation.
BOL is just the nameplate — on day one the plant has 100% of its datasheet energy, so the retention table can be read straight off the rating.
In reality: Nameplate is measured at the factory after formation; BOL is measured at the plant after transit, storage, installation and commissioning, and those months sit on the steepest part of the calendar-fade curve — an LFP fleet can arrive at its COD test a couple of percent down with every cell on model. BOL is whatever the acceptance test measures at the contract boundary under the contract conditions, and that number — not the rating — is the 100% the retention schedule divides by. Read the table off nameplate instead and the plant starts its warranty life already behind a curve it never agreed to.
Beginning of Life, in context.
The Grid-Scale BESS course covers beginning of life — and the rest of the system — from the ground up, the way it actually gets deployed.