Skip to main content
Library

Energy mass balanceCostingSolar

PV inverter conversion, clipping and levelized cost

Turns a PV inverter datasheet into AC energy delivered, conversion and clipping losses, and the levelized cost of the conversion step.

Photo of Open Jack

Open Jack

Founder · OpenJack

1

Cost of conversion

Levelized cost of conversion (LCO conversion)

Annualized inverter cost per MWh of AC energy delivered — a conversion adder, not a plant LCOE

…EUR/MWh

Levelized cost: first cost

Share of the conversion cost coming from the initial equipment purchase

…EUR/MWh

Levelized cost: replacement

Share coming from the mid-life inverter replacement

…EUR/MWh

Levelized cost: O&M

Share coming from fixed operation and maintenance

…EUR/MWh

Equipment price at the reference size of the selected class. Leave at 0 to use the published NREL 2024 benchmark of that class (0.0371 EUR/Wac central, 0.0682 EUR/Wac three-phase string), converted at the 2024 ECB average rate.

EUR/W

Inverter CAPEX (ex-factory equipment)

Equipment only — never the PV array; installation and indirects are added at project level

…EUR

Total annual cost

Annualized capital plus fixed O&M — no fuel term, the DC energy is priced upstream

…EUR/year

Inverter fixed O&M

Inverter share of plant O&M, taken as a ratio of its own CAPEX

…EUR/year
Scale effect on specific inverter cost

Log in to view this sensitivity chart.

Energy delivered

DC energy delivered by the array to this inverter over one year, after array-side losses. Enter it directly: no library calculator exposes DC-only electricity today (#429 already applies its own inverter loss and its BOS capex contains an inverter), so this input must not be chained in V1.

MWh/year

Annual AC energy delivered

DC input after conversion and clipping losses, capped at the AC nameplate over 8760 h

…MWh/year

Energy lost to clipping

Post-conversion energy the inverter cannot pass through at the chosen DC/AC ratio

…MWh/year

Energy lost to conversion

Electrical losses inside the inverter, excluding clipping

…MWh/year

Equivalent full-load hours (AC)

AC energy divided by AC nameplate — the annualization basis for the levelized cost

…h/year

AC capacity factor

Utilization of the AC nameplate over 8760 h — compare with the 24 % US utility-PV median

…%

Specific DC consumption per MWh of AC

Inverse of the delivered conversion ratio — DC energy needed for one MWh of AC

…MWh/MWh

Delivered conversion ratio

MWh_AC out per MWh_DC in, nameplate cap included — the coefficient to use on a DC-to-AC chaining link, once a DC-only upstream exists

…ratio
Utilization vs annual DC input

Log in to view this sensitivity chart.

Losses and sizing

Array DC nameplate divided by inverter AC nameplate. US median 1.34, 20th-80th percentiles 1.26-1.45. Drives clipping.

EN 50530 European efficiency or CEC efficiency from the datasheet. Default 98.1 % is the median of the inverter losses surveyed in IEA-PVPS T13-08.

Soiling, mismatch, wiring and thermal losses between module nameplate and inverter input. Sets the DC/AC ratio at which clipping starts. NREL standard is 14.1 %.

AC nameplate of the inverter. This is the cap on AC output and the basis of the CAPEX; 0.3 kW is a microinverter, 10 MW a containerised central unit.

Clipping loss

Share of post-conversion energy lost because the array peak exceeds the AC nameplate

…%

Conversion loss

DC-to-AC electrical loss at the weighted efficiency, before clipping

…%

Total DC-to-AC loss

Conversion and clipping combined, as a share of the DC energy entering the inverter

…%

Clipping onset DC/AC ratio

DC/AC ratio below which no clipping occurs, at the DC losses entered

…ratio

Reference clipping onset (NREL derate)

Onset at the NREL standard 85.9 % pre-inverter DC derate, for comparison

…ratio

Average DC load ratio

Annual-average DC power divided by DC nameplate — where on the efficiency curve the unit lives

…ratio
Clipping and cost vs DC/AC ratio

Log in to view this sensitivity chart.

Sensitivity to weighted efficiency

Log in to view this sensitivity chart.

Equipment and benchmark

Sets the reference size and scale exponent of the cost curve, and the published NREL benchmark shown next to your price.

Scaled specific CAPEX

Inverter price per W of AC nameplate after the class scale curve and clamps

…EUR/W

Scaled specific CAPEX in 2024 USD

Same figure in the currency the NREL benchmark is published in

…USD/W

Class benchmark specific CAPEX

Published NREL 2024 benchmark for the selected inverter class, to sit next to your own price

…EUR/W

Entered price vs class benchmark

100 % means the entered price sits exactly on the NREL benchmark for that class

…%

Finance and replacement

Real equity discount rate used for the CRF and for discounting the replacement. NREL 2024 utility PV benchmark uses 5.9 %.

Cost-recovery period. NREL ATB 2024 estimates utility PV energy over a 30-year lifetime.

years

Year of the single mid-life inverter replacement. IEA-PVPS reports a technical life of 10-15 years and an MTBF around 11-12 years.

Capital recovery factor

Annualizes capital at the real discount rate over the plant lifetime

…1/year

Annualized CAPEX

Capital commitment converted to an equivalent annual payment

…EUR/year

Mid-life replacement, discounted

Present value of one inverter replacement at the year entered

…EUR

Total capital commitment (first cost + replacement PV)

What the CRF is applied to over the plant lifetime

…EUR
Cost of the replacement year

Log in to view this sensitivity chart.

About

Calculator context

Takes one grid-connected PV inverter, central or three-phase string, and answers the two questions a designer actually has about it: how much of the array's DC energy reaches the grid as AC, and what that conversion step costs per MWh delivered.

The AC nameplate is entered in kW, the way every datasheet states it. Conversion follows the weighted efficiency of the datasheet (EN 50530 European or CEC), and clipping follows the DC/AC ratio: nothing is clipped until the derated array peak passes the AC cap, which is why the onset is computed from the pre-inverter DC losses rather than assumed. Above that onset the annual clipped fraction comes from a two-parameter correlation fitted to the published contiguous-US modelling of Micheli et al. (2024), reproducing 1.0 % at ILR 1.34 with 10 % DC losses and 4.5 % at ILR 1.45 with 5 % losses. The AC energy is then capped, as a real inverter caps it, at the nameplate over 8760 hours.

On the money side, the equipment price is anchored on the NREL 2024 benchmark inverter line, converted to euros per W of AC nameplate at the ECB 2024 average rate, and one mid-life replacement is discounted and annualized alongside the first cost. There is no fuel term: the DC electricity is priced by whatever produces it, and charging it here would count the array twice.

Scope, stated plainly. The CAPEX covers the inverter and only the inverter. This asset is meant to sit behind a module-level DC source, never behind an all-in plant calculator whose cost already includes its own conversion stage. On such a link both sides carry the electricity commodity, so the coefficient to use is the delivered conversion ratio this calculator reports, never an arbitrary partial coefficient.

Model

93 variables — inputs, calculations and outputs, with their dependencies.

93 variables shown of 93
VariableValueUnitDepends on
1——
1000kW—
1.34ratio—
98.1%—
2150MWh/year—
14.1%—
0EUR/W—
12years—
0.059ratio—
30years—
VariableFormulaUnitDepends on
if((<=)+(<)+(<=)+(>)+(<)+(<)+(<)+(>=)+(<)+(>)+(<)+(<)+(!=)*(!=)>,,)bool
clamp(-*,,)ratio
max(-,)ratio
min(,*(^)*)ratio
**MW
/max(,)MW
clamp((-*-/max(,))/max(,),,)ratio
clamp((-*-/max(,))/max(,),,)ratio
clamp(+*(/max(,)-),,)ratio
/max(,)%
**(-)ratio
*MWh/year
*MWh/year
**MWh/year
**MWh/year
if(==,,)kW
if(==,,)ratio
if(<=,,)EUR/W
*(/max(,))^(-)EUR/W
/max((+)^,)ratio
VariableFormulaUnitDepends on
/max(,)ratio
/max(,)ratio
/max(,)ratio
-%
min(,)MWh/year
(-)/max(,)*%
/max(,)ratio
-*%
max(-,)MWh/year
*(-*)MWh/year
/max(*,)h/year
/max(,)*%
/max(,)MWh/MWh
if(==,,)EUR/W
/max(,)*%
clamp(,,)EUR/W
*USD/W
**EUR
*EUR
+EUR
if(<,/max(,),(*(+)^)/max((+)^-,))1/year
*EUR/year
*EUR/year
+EUR/year
/max(,)EUR/MWh
*/max(,)EUR/MWh
*/max(,)EUR/MWh
/max(,)EUR/MWh

Assumptions

68 assumptions used in the calculations

  • Prevents division by zero and NaN propagation in the DSL guards.

    Market range Not applicable (numerical parameter).

    0.000001
    Numerical stability constant for division guards.

  • Neutral element used instead of inline literals, so every number in the DSL is a named, sourced constant.

    Market range Exact.

    1
    Modelling convention.

  • Neutral element for the validity guards and for the clipping headroom floor.

    Market range Exact.

    0
    Modelling convention.

  • Percent base used to convert fractions to percent and back.

    Market range Exact.

    100%
    Unit convention.

  • Converts a percent input (0-100) into the fraction the physics uses.

    Market range Exact.

    0.01
    Unit convention (OpenJack: percent inputs are 0-100, never 0-1).

  • Time basis for the AC energy cap and for the capacity factor.

    Market range 8760 (non-leap year); 8784 (leap year).

    8760h/year
    Calendar-year hours convention.

  • Converts the kW nameplate times 8760 h into MWh for the AC energy cap.

    Market range Exact.

    0.001MWh/kWh
    SI unit conversion.

  • Converts the kW nameplate into W so the EUR/W specific CAPEX yields euros.

    Market range Exact.

    1000W/kW
    SI unit conversion.

  • Converts the kW nameplate into MW for the DC load ratio and the full-load hours.

    Market range Exact.

    0.001MW/kW
    SI unit conversion.

  • Integer code for the central (utility-scale) inverter class.

    Market range Not applicable (selector code).

    1
    Selector encoding (the DSL carries no strings).

  • Integer code for the three-phase string (commercial) inverter class.

    Market range Not applicable (selector code).

    2
    Selector encoding (the DSL carries no strings).

  • Published benchmark price for a utility-scale central inverter, per W of AC nameplate.

    0.0371EUR/W
    $0.03/Wdc (NREL 2024 MMP utility breakdown) / 1.0824 USD per EUR (ECB 2024 average) x ILR 1.34.

  • Published benchmark price for a commercial three-phase string inverter, per W of AC nameplate.

    0.0682EUR/W
    $0.06/Wdc (NREL 2024 MMP commercial breakdown) / 1.0824 USD per EUR x ILR 1.23 (Table A-7).

  • Currency conversion applied to the NREL benchmark, which is published in 2024 USD.

    1.0824USD/EUR
    Annual average of the 256 published business-day USD reference rates for 2024.

  • Reference size at which the central-class benchmark price applies; the scale curve pivots here.

    1000kW
    Declared nominal case of this calculator, at the scale of a utility central inverter block.

  • Reference size at which the three-phase string benchmark price applies.

    100kW
    Typical nameplate of the three-phase string inverters used in the NREL commercial benchmark.

  • Cost-capacity exponent inside the central class: specific cost falls as size^(exponent-1).

    0.9
    Screening value for modular power electronics, where cost is close to linear in rated power.

  • Cost-capacity exponent inside the string class, flatter still because units are simply added.

    0.95
    Screening value for modular string inverters.

  • Lower guardrail on the scaled specific CAPEX.

    Market range Not a market quote; a guardrail.

    0.015EUR/W
    Modelling clamp set below the cheapest published class benchmark.

  • Upper guardrail on the scaled specific CAPEX.

    Market range Not a market quote; a guardrail.

    0.15EUR/W
    Modelling clamp set above the most expensive published class benchmark.

  • Fixed O&M attributed to the inverter, as a fraction of its own CAPEX per year.

    0.0091/year
    NREL ATB system-related fixed O&M to CAPEX ratio of 0.9:100 for 2023.

  • Standard DC-side derate used to place the reference clipping onset.

    0.859
    NREL one-axis tracker pre-inverter derate of 85.9 %, i.e. 14.1 % DC losses.

  • Floor on the DC derate factor, so an extreme DC-loss entry cannot send the clipping onset to infinity.

    Market range Not a market quote; a guardrail.

    0.5
    Modelling guardrail.

  • Coefficient of the annual clipping correlation, in percent of post-conversion energy.

    Market range Reproduces 1.00 % at ILR 1.34 with 10 % DC losses and 4.51 % at ILR 1.45 with 5 % DC losses.

    56.2%
    Fitted, with the exponent, to the two quantified clipping anchors of Micheli et al. (2024), each at its own DC-loss assumption.

  • Exponent of the annual clipping correlation on the DC/AC headroom above the onset.

    Market range Clipping rises steeply once the derated array peak passes the AC cap.

    2.733
    Fitted jointly with the coefficient to the same two published anchors.

  • Constant term of the PVWatts v5 inverter part-load efficiency curve.

    Market range Published model coefficient.

    0.9858
    PVWatts v5: eta = (eta_nom/eta_ref) * (-0.0162 z - 0.0059/z + 0.9858), z = Pdc/Pdc0.

  • Linear term of the PVWatts v5 part-load efficiency curve.

    Market range Published model coefficient.

    0.0162
    PVWatts v5 coefficient on z.

  • Inverse term of the PVWatts v5 curve, which is what makes efficiency collapse at very low load.

    Market range Published model coefficient.

    0.0059
    PVWatts v5 coefficient on 1/z.

  • Normalisation of the PVWatts curve, chosen so the shape equals one at full load.

    Market range Published model coefficient (default 0.9637).

    0.9637
    PVWatts v5 reference efficiency eta_inv_ref.

  • Load point the weighted efficiency is centred on, used as the reference of the part-load correction.

    Market range 0.5 (Euro weighting) or 0.75 (CEC weighting, weight 0.53).

    0.5
    EN 50530 European efficiency puts a weight of 0.48 on the 50 % load point, the largest of the six.

  • Strength of the part-load correction applied on top of the weighted efficiency. Zero by default.

    Market range 0 (default, no double count) to 1 (full PVWatts shape correction).

    0
    Kept at zero because the surveyed inverter losses are annual plant figures that already embed part-load operation; raising it would double-count.

  • Floor on the part-load shape, since the 1/z term diverges as load approaches zero.

    Market range Not a market quote; a guardrail.

    0.5
    Modelling guardrail.

  • Cap on the part-load shape; the PVWatts curve peaks at about 1.0026 near 60 % load.

    Market range Curve maximum is 1.0026 at z = 0.60.

    1.01
    Modelling guardrail just above the analytic maximum of the curve.

  • Lower clamp on the part-load penalty, so the correction can never improve on the weighted efficiency.

    Market range 1 by definition.

    1
    Modelling guardrail.

  • Upper clamp on the part-load penalty.

    Market range Not a market quote; a guardrail.

    1.5
    Modelling guardrail.

  • Default DC/AC ratio of the input, and the US median.

    Market range 1.26 (20th percentile) to 1.45 (80th percentile); 1.10 in 2010.

    1.34ratio
    NREL ATB 2024 representative utility plant: 100 MWdc / 74.6 MWac, one-axis tracking.

  • Low end of the US ILR distribution, used to frame the input range.

    1.26ratio
    20th percentile of US utility-scale projects.

  • High end of the US ILR distribution, and the anchor of the high-clipping scenario.

    1.45ratio
    80th percentile of US utility-scale projects.

  • Historical low of the fleet ILR, which is why the input allows ratios down to 1.

    Market range 1.10 (2010) rising to 1.34 (2019-2024).

    1.1ratio
    NREL utility PV LCOE assumptions, one-axis tracker ILR in 2010.

  • DC/AC ratio below which clipping cannot occur at the NREL standard DC derate. Computed in the model, not entered.

    Market range 1.05 (5 % DC losses) to 1.18 (15 % DC losses).

    1.164ratio
    1 / 0.859, with 0.859 the NREL pre-inverter DC derate.

  • Published clipping loss for tracking systems at ILR 1.34, contiguous-US average.

    Market range 0.6 % (single-axis tracking) to 1.4 % (fixed tilt) at the same ILR.

    0.6%
    5-minute NSRDB modelling across the contiguous US.

  • Published clipping loss for fixed-tilt systems at ILR 1.34, contiguous-US average.

    1.4%
    5-minute NSRDB modelling across the contiguous US.

  • Published clipping loss in the high-clipping scenario, ILR 1.45 with only 5 % DC losses.

    Market range Stated as between 4 % and 5 %.

    4.5%
    Second anchor of the fitted correlation.

  • Lowest annual inverter loss among the surveyed PV financial models.

    1.1%
    Seven surveyed models: 1.1, 1.6, 1.7, 1.9, 2.0, 2.2, 3.2 %.

  • Median annual inverter loss of the surveyed models, which sets the default efficiency.

    1.9%
    Median of the seven surveyed values.

  • Highest annual inverter loss among the surveyed models; frames the top of the loss output range.

    Market range Uncertainty stated as plus or minus 1.0 to 2.0 points.

    3.2%
    Seven surveyed models.

  • Default weighted conversion efficiency of the input.

    98.1%
    100 minus the 1.9 % median inverter loss of the surveyed financial models.

  • Inverter efficiency assumed inside the NREL ATB capacity-factor model, for comparison.

    96%
    ATB 2024 utility PV: 96 % in 2021 and in the Conservative Scenario, 98 % in Moderate and Advanced.

  • Low end of the inverter technical life, which frames the replacement-year input.

    10years
    IEA-PVPS: inverter life considered to be between 10 and 15 years.

  • High end of the inverter technical life.

    15years
    IEA-PVPS: inverter life considered to be between 10 and 15 years.

  • Field mean time between failures reported across climates, the empirical basis for the default replacement year.

    11.5years
    IEA-PVPS: around 11-12 years across past studies.

  • Default replacement year of the input.

    12years
    Field MTBF of 11-12 years; IEA-PVPS advises anticipating replacement roughly halfway through plant life.

  • Default cost-recovery period of the input.

    30years
    AI

  • Default real discount rate of the input.

    0.059ratio
    NREL utility PV LCOE assumptions: real equity discount rate 5.9 % for 2023-2024.

  • Whole-system fixed O&M, used only as an order-of-magnitude check on the inverter-only O&M.

    22EUR/kW/year
    NREL ATB 2024: $22/kWac-yr in 2023, split system 14.4, property 5.4, administration 2.4.

  • Lowest US resource-class AC capacity factor, framing the capacity-factor output.

    0.214ratio
    NREL ATB 2024, Class 10 (annual GHI below 3.75).

  • Highest US resource-class AC capacity factor.

    0.34ratio
    NREL ATB 2024, Class 1 (annual GHI above 5.75).

  • Cumulative median AC capacity factor of the installed US utility-scale fleet; the declared nominal case reproduces it.

    0.24ratio
    Bolinger et al. (2023) via the ATB: 24 % for projects installed 2007-2021, project range 9-35 %.

  • AC nameplate of the declared nominal case.

    1000kW
    1000 kWac central inverter, the size the default benchmark price is anchored on.

  • Annual DC input of the declared nominal case, chosen so the result lands on the measured US median.

    2150MWh/year
    1000 kWac x 8760 h x 24 % AC capacity factor, back-solved through 98.1 % efficiency and 0.49 % clipping.

  • Published utility inverter price before currency and Wdc-to-Wac conversion.

    Market range Stable around $0.03/Wdc since 2022.

    0.03USD/W
    NREL 2024 MMP utility PV cost breakdown, 2024 USD.

  • Published commercial inverter price before currency and Wdc-to-Wac conversion.

    Market range Roughly twice the utility figure per Wdc.

    0.06USD/W
    NREL 2024 MMP commercial PV cost breakdown, 2024 USD.

  • Dominant weight of the EN 50530 European efficiency, which is why 50 % load is the reference point.

    Market range Weights are fixed by the standard.

    0.48
    European efficiency = 0.03 h5 + 0.06 h10 + 0.13 h20 + 0.10 h30 + 0.48 h50 + 0.20 h100.

  • Dominant weight of the CEC efficiency, which centres higher than the European weighting.

    Market range Weights are fixed by the protocol.

    0.53
    CEC efficiency = 0.04 h10 + 0.05 h20 + 0.12 h30 + 0.21 h50 + 0.53 h75 + 0.05 h100.

  • The CAPEX covers the inverter and nothing else. It never includes the PV array, its mounting or its wiring.

    Market range Not applicable (scope rule).

    —
    Scope decision: this asset models conversion alone, so its cost boundary stops at the inverter.

  • This asset must not be chained to any solar calculator published today: an all-in plant (#242, #243) already contains the conversion in its CAPEX, and the only module-level calculator (#429) already outputs AC and carries an inverter in its BOS.

    Market range Not applicable (chaining rule).

    —
    Chaining doctrine, revised 2026-09-03 after checking #429: its inverter_loss_frac enters its performance ratio and its bos_capex_ref covers the inverter, so chaining it here double-counts both the conversion loss and the inverter CAPEX. Until a DC-only source exists, this calculator is a component comparator, not a project node.

  • On a DC-to-AC link the coefficient is the delivered conversion ratio, never an arbitrary partial coefficient.

    Market range Not applicable (chaining rule).

    —
    The reference set has a single electricity commodity, so inlet and outlet share it; the coefficient carries the physics.

  • Clipping is closed annually by a fitted correlation, not integrated from a time series.

    Market range Real clipping is site and mounting specific.

    —
    A unit calculator has no hourly data; the correlation reproduces published fleet-average results.

Sources

6 external sources