Levelised Cost of Electricity for Solar, Calculated Properly

Payback period answers a cash-flow question: when do I get my money back? Levelised cost of electricity answers a different and more useful one: what does a kilowatt-hour from this asset actually cost me over its life? Because LCOE is expressed in currency per kilowatt-hour, it can be compared directly against a retail tariff, a power purchase agreement price or the LCOE of any other generator.

It is also easy to get wrong, and the wrong versions almost always flatter solar.

The definition

LCOE is the constant price per kilowatt-hour that, over the asset's life, exactly balances the present value of all costs against the present value of all generation.

`` LCOE = Σ [ (C_t + O_t) / (1+r)^t ] / Σ [ E_t / (1+r)^t ] ``

where C_t is capital expenditure in year t, O_t is operating expenditure in year t, E_t is energy generated in year t, and r is the discount rate.

The critical structural point, and the one most spreadsheets miss: the energy in the denominator must be discounted too. A kilowatt-hour delivered in year 24 is not worth the same as one delivered in year 1, for the same reason that a euro is not. Omitting the discount on energy understates LCOE substantially — by roughly 30 to 45 percent at a 5 percent discount rate over 25 years. If a published LCOE looks implausibly low, this is the first place to check.

The inputs, and what each one really is

Capital cost. Everything paid to get to commissioning: modules, inverters, mounting, cabling, protection, labour, scaffolding, design, grid application fees, and any structural works. Deduct grants and tax credits in the year they are actually received, not at signature.

Operating cost. Cleaning, inspection, monitoring subscriptions, insurance, and an allowance for fault-finding. For residential systems 0.5 to 1.0 percent of capital cost per year is a defensible starting point; large ground-mounted plants run lower per kilowatt but add land lease and security.

Replacement cost. String inverters typically need replacement once in a 25-year life, commonly between years 10 and 15. Budget for it explicitly in the year it falls, discounted. Ignoring inverter replacement is the second most common error after undiscounted energy.

Annual generation. From an hourly model, not a rule of thumb. Then apply degradation.

Degradation. Modern modules lose roughly 0.4 to 0.55 percent per year after an initial first-year drop of 1 to 2 percent for silicon. Warranty terms typically guarantee 85 to 88 percent of nameplate at year 25, which is a floor, not an expectation. See module degradation and warranties.

Lifetime. 25 years is the conventional modelling life for a residential system. Physically, well-built arrays run considerably longer; financially, the discounted contribution of years 26 to 35 is small.

Discount rate. The most consequential and least discussed input. For a household, the honest rate is the return foregone — the mortgage rate if the money would otherwise reduce debt, or a realistic savings return. For a commercial project, the weighted average cost of capital. Use a real rate against real (inflation-adjusted) cash flows, or a nominal rate against nominal ones, and never mix the two.

A worked example

A 6 kWp residential system in central Europe.

• Capital cost after grant: 7,800 EUR
• Year-1 generation: 6,900 kWh (1,150 kWh/kWp)
• Degradation: 2 percent in year 1, 0.5 percent per year thereafter
• Operating cost: 70 EUR/yr, rising with inflation
• Inverter replacement: 1,200 EUR in year 13
• Life: 25 years
• Real discount rate: 4 percent

Discounted lifetime generation comes to about 105,000 kWh, against undiscounted generation of roughly 162,000 kWh — the size of the gap illustrates the point above. Discounted lifetime cost, including the year-13 replacement, is about 9,000 EUR.

`` LCOE = 9,000 / 105,000 = 0.086 EUR/kWh ``

Against a retail import price of 0.28 EUR/kWh, that is a comfortable margin. Against an export price of 0.07 EUR/kWh, exported energy is generated at a loss on a levelised basis. Both statements are true simultaneously, and together they explain why self-consumption dominates residential solar economics.

The table is the honest summary of how much a single unmeasured assumption moves the answer. Any LCOE quoted without its discount rate is not a number, it is a mood.

Comparing LCOE against the right benchmark

A rooftop LCOE should be compared against the retail tariff, because rooftop generation displaces retail purchases. A utility-scale LCOE should be compared against the wholesale price, because that is what it competes with. Comparing rooftop LCOE against utility-scale LCOE and concluding rooftop is expensive misses that they are selling into different markets at different prices.

The correct comparison for a household is: is my LCOE below the price I would otherwise pay, weighted by how much of my generation I actually consume?

`` Blended value = SC × P_import + (1 − SC) × P_export ``

If blended value exceeds LCOE, the project is sound. With 40 percent self-consumption in the example above, blended value is 0.40 × 0.28 + 0.60 × 0.07 = 0.154 EUR/kWh against an LCOE of 0.086 — a healthy margin. Push self-consumption to 60 percent and blended value rises to 0.196.

What LCOE deliberately excludes

LCOE says nothing about timing of value, resilience during outages, the option value of adding storage later, roof lifetime interactions, or the effect on property value. It is a cost metric, not a decision. Use it to compare generation options on a common basis, then bring in the factors it omits separately.

It also assumes the generation is sellable at a single price. Under time-varying tariffs, a value-adjusted metric — sometimes called LACE, the levelised avoided cost of energy — is the more informative counterpart.

Common mistakes

• Not discounting energy. Understates LCOE by a third or more.
• Omitting inverter replacement. Understates by roughly 0.008 to 0.015 EUR/kWh on a residential system.
• Using nameplate generation with no degradation. Overstates lifetime output by 6 to 8 percent.
• Mixing real and nominal. Escalating operating costs with inflation while using a real discount rate double-counts inflation.
• Applying a zero discount rate. Legitimate only if the money genuinely has no alternative use, which is almost never true.
• Deducting a tax credit at year zero when it arrives in year two. Small, but it is the kind of error that travels with the rest.

Frequently asked questions

What is a typical LCOE for residential rooftop solar? In Europe, 0.06 to 0.12 EUR/kWh for a well-priced system in a decent location at a 3 to 5 percent real discount rate. Northern latitudes with low specific yield and high labour costs sit at the upper end; southern installations with strong irradiation sit near the bottom.

Is LCOE better than payback period? They answer different questions. Payback is a liquidity measure and is easy to communicate; LCOE is a cost measure and is comparable across technologies. A serious assessment computes both, plus net present value, and checks that all three tell a consistent story.

How does a battery affect LCOE? Storage raises the cost per delivered kilowatt-hour, because it adds capital and losses without adding generation. Its justification is that it raises the value of each kilowatt-hour by moving it into a more expensive hour. Treat storage in the value calculation, not by hoping the LCOE improves.

What discount rate should a household use? The rate on the debt the money could otherwise repay, or the realistic after-tax return on the savings it comes from. For most households that is 3 to 6 percent in real terms. Present a range rather than a single figure.

References

• IEA and NEA, Projected Costs of Generating Electricity
• Fraunhofer ISE, Levelized Cost of Electricity: Renewable Energy Technologies
• IRENA, Renewable Power Generation Costs
• NREL, Annual Technology Baseline documentation on LCOE methodology