Solar Panel Efficiency vs. Real-World Output: What Actually Reaches the Meter

Quick Answer

Why This Matters for EPCs

Every panel datasheet leads with one number: efficiency at Standard Test Conditions (STC). It's the number homeowners ask about, the number that sells premium panels, and the number that is least representative of what a system will actually produce once it's bolted to a roof or staked into the ground.

The gap between STC efficiency and delivered energy isn't a rounding error. Stack temperature, degradation, soiling, wiring, mismatch, and inverter clipping together, and a "22% efficient" panel can deliver the real-world equivalent of 16-17% in year one, before 25 years of gradual decline are even factored in. None of this is hidden or disputed. NREL publishes the loss model. Sandia publishes the temperature physics. The problem is that almost nothing written about "solar panel efficiency" puts these pieces together into one number an EPC can actually use to spec a project.

This piece does that. It walks through every real-world derating factor with the underlying data, then applies the full stack to three real project types EPCs quote every week, so the output isn't "efficiency matters more when space is limited," it's an actual delivered-kWh comparison with the math shown.

1. What the Datasheet Number Actually Measures

Solar panel efficiency is the percentage of sunlight energy a panel converts into usable electricity, measured under Standard Test Conditions: 25°C cell temperature, 1000 W/m² irradiance, and an air mass of 1.5 (a standardized approximation of sunlight passing through the atmosphere at a specific angle).

Efficiency is calculated as:

Efficiency (%) = (Rated power output at STC ÷ panel area in m²) ÷ 1000 W/m² × 100

A 2 m² panel rated at 400 W under STC is 20% efficient. That's it. It's a lab measurement, not a field forecast.

Two layers determine that number:

• Cell efficiency: how much of the light hitting the silicon actually converts to current, determined by material purity, cell architecture, and surface treatment.
• Panel efficiency: cell efficiency after accounting for the gaps between cells, the frame, busbars, and the glass/encapsulant losses that don't exist at the bare-cell level. Panel efficiency is always somewhat lower than the best cell efficiency inside it.

Typical STC efficiency by cell technology, current market:

These bands shift every year as manufacturing improves. What doesn't shift is the underlying problem with using this number alone to spec a project: it describes performance in a condition that essentially never occurs on a real roof or field.

2. The First Gap: NREL's Own System Loss Model

The National Renewable Energy Laboratory's PVWatts calculator, the most widely used solar production estimator in the US, doesn't apply STC efficiency directly to a production forecast. It applies STC efficiency and then subtracts a documented stack of system losses, because NREL's own modeling accounts for the fact that STC output and delivered output are different things.

PVWatts' default total system loss is 14%, and it is not a flat 14% subtracted once. It's built from individual loss factors combined multiplicatively (each loss is applied to what's left after the previous one, not summed):

This is before temperature is even applied, and before the DC-to-AC inverter conversion loss is separately modeled. It's also before the panel's own real-world operating temperature is factored in, which is the single largest of all the derating factors and gets its own section below because it deserves one.

The point of walking through this list isn't the 14% figure itself, defaults are defaults, real projects deviate from them constantly. It's that NREL's own reference tool treats "efficiency at STC" as a starting input, not an output. Any spec conversation that stops at the datasheet number is skipping the step the US government's own solar modeling software doesn't skip.

3. Temperature: The Largest Real-World Derate, and the One Most Content Skips

Here's what almost nothing written about panel efficiency mentions: heat costs more real-world output than any other single factor on this list, including shading, soiling, and wiring combined, and it applies to every single panel, every single day, regardless of installation quality.

The mechanism

Every crystalline silicon panel has a published temperature coefficient of power (usually written as %/°C), which describes how much power output drops for every degree Celsius above the 25°C STC reference temperature. Sandia National Laboratories' PV Performance Modeling Collaborative (PVPMC) formalizes this in the Sandia Array Performance Model, the reference framework the solar industry uses to model temperature-corrected output.

Typical ranges by technology, industry-wide:

A lower magnitude (closer to zero) is better, it means the panel loses less power as it heats up.

Why this matters more than the STC number in hot climates

Panels don't operate at 25°C. On a clear, hot day, a rooftop panel's cell temperature routinely runs 20-40°C above ambient air temperature, meaning a 35°C (95°F) day in Phoenix or Dallas can put cell temperatures at 55-70°C.

Take a realistic mid-summer scenario: 65°C cell temperature, a panel with a typical -0.35%/°C coefficient.

• Temperature differential: 65°C - 25°C = 40°C
• Power loss: 40 × 0.35% = 14%
• Actual output: 86% of rated power, from heat alone, before any other loss in the stack is applied

Now compare two panels on the same project, identical STC rating, different temperature coefficients:

That 5.6-point gap is real, year-round, compounding daily, in every hot-climate market, and it exists whether or not the STC efficiency numbers on the two datasheets are identical. Two panels rated at the same 21.5% STC efficiency can deliver meaningfully different real-world output in Phoenix specifically because of this number, and this number alone rarely makes it into a sales conversation that stops at the headline efficiency percentage.

This is also why the same panel model performs differently by region. A -0.40%/°C panel in coastal San Diego (mild temperatures nearly year-round) loses far less to heat than the identical panel in Las Vegas. Efficiency comparisons that ignore climate are comparing lab numbers, not field numbers.

4. Degradation: What "22% Efficient" Means in Year 25, Not Year 1

Every panel loses a small amount of output every year it operates, a combination of light-induced degradation, potential-induced degradation, UV exposure, thermal cycling, and micro-cracking. This is separate from, and additive to, the temperature loss above.

The most cited field-data review on this comes from Dirk Jordan and Sarah Kurtz at NREL, first published in Progress in Photovoltaics and later expanded into a compendium covering data from thousands of field-measured systems. Their finding, based on median degradation across the dataset: crystalline silicon panels degrade at a median rate of roughly 0.5-0.6% per year.

That means a panel doesn't hold 100% of its rated output for 25 years and then fall off a cliff. It declines steadily:

This is, not coincidentally, close to what manufacturer performance warranties already guarantee (commonly around 87.4-90% of nameplate at year 25 or 30 on current Tier 1 datasheets), which is a useful sanity check: the field data and the warranty language broadly agree. Where they diverge is technology-specific. NREL's later compendium work found HIT (heterojunction) and microcrystalline silicon technologies degrading closer to 1%/year, more in line with thin-film behavior than standard crystalline silicon, an important distinction when a "higher efficiency, premium technology" panel is being evaluated for a 25-year investment and not just first-year output.

The EPC-relevant point: two panels with identical year-one STC efficiency can diverge by several percentage points of cumulative output by year 15 purely on degradation rate, a number that almost never appears next to the efficiency percentage in a sales conversation, and one that compounds with the temperature coefficient above rather than replacing it.

5. Bifacial Gain: Real, But Entirely Dependent on the Site, Not the Panel

Bifacial modules generate power from both the front and rear surface, using a transparent backsheet or glass-glass construction to let reflected light reach the rear cells. The additional energy this produces is called bifacial gain, and it is one of the most commonly overstated figures in solar marketing, not because the technology doesn't work, but because the gain is a property of the installation, not a fixed spec on the panel.

Field data (drawing on NREL, Sandia, and IEA PVPS measurement programs, aggregated across multiple published studies) shows gain varying by an order of magnitude depending on three factors:

• Ground/surface albedo (reflectivity): fresh snow reflects roughly 80-95% of incident light, white membrane roofing roughly 70-80%, concrete 20-30%, grass 15-25%, dark shingle roofing under 10%
• Mounting height: more clearance between the panel's rear surface and the ground/roof allows more reflected light to reach the rear cells
• Tracking vs. fixed-tilt: single-axis trackers expose more of the rear surface to reflected light across the day than a fixed-tilt array

Realistic gain ranges by installation type:

The practical implication for an EPC quoting a project: the same bifacial module can be a strong economic case on one roof and a wasted premium on another. A residential flush-mount job on dark shingles is very unlikely to recover a bifacial price premium (commonly 3-8% over monofacial equivalents). The identical panel on an elevated C&I rack over a white membrane roof is a materially different calculation. This is a design decision, not a panel-spec decision, and it's the kind of determination that depends on actually knowing the roof surface and mounting height for the specific project, not a general efficiency rating.

6. The Factor Almost No "Efficiency" Content Mentions: DC:AC Ratio and Clipping

This is the piece of the puzzle that's genuinely missing from most solar panel efficiency content, on any site, in any market, and it directly interacts with everything above.

Most PV systems are designed with more DC capacity (panel wattage) than the inverter's rated AC output, a deliberate design choice called the inverter loading ratio (ILR), or DC:AC ratio. Because inverters cost more per kW than panels, and because panels rarely operate at their peak rating simultaneously across an entire array, oversizing the DC side and letting the inverter "clip" the rare peak moments is usually the more economical design.

NREL's own published research (Denholm et al., 2017) found a typical utility-scale ILR around 1.3, with more recent NREL bottom-up cost analysis (2021-2022) putting the optimal range closer to 1.28-1.34. EIA-observed data for individual systems shows a slightly wider real-world spread, roughly 1.13-1.30.

Where this intersects directly with everything above:

• Higher-efficiency panels increase DC capacity per roof area, which can quietly push a project's ILR higher than intended if the inverter wasn't resized to match, meaning more of the panel's gain gets clipped away rather than delivered.
• Bifacial panels have a measured, real-world DC output higher than their front-side nameplate rating implies (that's the entire point of bifacial gain), which means a system designed to a monofacial-equivalent ILR target will run a higher effective ILR than modeled, and clip more than expected, unless the design accounts for it. NREL's own PV modeling guidance notes explicitly that the optimal ILR for bifacial systems should generally run lower than for monofacial systems for exactly this reason.
• Clipping loss is nonlinear. Published clipping-loss modeling shows minimal loss around an ILR of 1.25, but losses climbing sharply at higher ratios, one widely cited minute-resolution study found roughly 16% of potential annual generation lost at an ILR of 2.0.

The practical upshot: a project spec'd around "high efficiency panels for maximum output" without revisiting inverter sizing can end up losing part of that gain right back to clipping, silently, with no line item anywhere calling it out. This is a design-software problem as much as a panel-spec problem, which is exactly why it's covered in the workflow section below.

7. Putting the Stack Together: What "22% Efficient" Actually Delivers

None of the factors above apply in isolation. A real project stacks all of them. Here's what that stack looks like when the individual factors above are combined (multiplicatively, consistent with how NREL's own PVWatts model combines them) for a representative hot-climate rooftop scenario:

Starting point: 22% STC-rated monocrystalline panel

Effective delivered efficiency in year one: roughly 15.5-16%, from a panel rated at 22%. That's before 25 years of degradation at 0.5-0.6% annually are applied on top.

This isn't a criticism of the panel or the manufacturer, every panel on the market carries some version of this gap, because STC is a lab condition and every installation is a field condition. It's the reason two projects using "the same 22% efficient panel" can produce meaningfully different actual output: the gap between the two isn't the panel, it's the site, the climate, the mounting, and the system design around it.

8. Three Real Projects, Worked Through the Full Stack

Illustrative modeling below, using the published loss factors and ranges cited throughout this piece. Real project output depends on the specific site's shading, orientation, and equipment, and should be run through full production modeling before being quoted, not estimated from these examples directly.

Residential rooftop, Phoenix, AZ (1,400 sq ft usable roof, dark asphalt shingle, flush-mount)

• Panel choice: monocrystalline PERC/TOPCon, prioritizing temperature coefficient over marginal STC efficiency gains, given Phoenix's extreme heat exposure
• Bifacial: not a meaningful factor here (dark shingle, flush-mount, negligible rear-side gain)
• Dominant loss factor: temperature. Phoenix summer cell temperatures regularly exceed 65-70°C. A panel with a -0.40%/°C coefficient versus one at -0.28%/°C represents a real, multi-percentage-point difference in delivered summer output, precisely the months when the utility rate and demand charges are highest.
• ILR consideration: residential string inverters are typically sized closer to a 1.1-1.2 ILR given panel-count granularity on a constrained roof; less clipping risk than a utility design, but still worth checking against the specific inverter's rated capacity once the panel count is finalized.
• Decision framework: on this roof type, temperature coefficient and degradation rate matter more to 25-year output than an extra half-point of STC efficiency. The panel that "wins" the datasheet efficiency comparison isn't automatically the one that wins the Phoenix production comparison.

C&I flat roof, Dallas, TX (40,000 sq ft warehouse roof, white TPO membrane, elevated ballasted racking)

• Panel choice: this is the scenario where bifacial modules earn their premium. White TPO membrane is high-albedo, and elevated ballasted racking provides real rear-side clearance, both conditions that push bifacial gain toward the 10-20% range rather than the 2-5% seen on residential dark-shingle roofs.
• Bifacial: a legitimate case for the 3-8% cost premium, given the albedo and mounting height both work in its favor, unlike the Phoenix residential scenario above.
• Dominant loss factors: temperature remains significant (Dallas summers are comparably harsh to Phoenix), and bifacial gain works in the project's favor here, partially offsetting it.
• ILR consideration: this is exactly the situation flagged in Section 6. Because bifacial modules deliver real output above their front-side nameplate rating, an inverter sized to a standard monofacial ILR target risks clipping more than modeled. Design software needs to size the inverter to the effective DC output including bifacial gain, not just the front-side nameplate.
• Decision framework: bifacial plus a slightly more conservative ILR than the monofacial default is the likely optimal spec here, a genuinely different conclusion than the residential case above, even though both are "high efficiency panel" decisions on the surface.

Utility-scale ground-mount, West Texas (50 MW, single-axis tracker, light soil/caliche surface)

• Panel choice: bifacial is close to default at this scale and site type in 2026; the real design questions are tracker geometry, row spacing, and ILR, not whether to use bifacial at all.
• Bifacial: single-axis tracking plus moderate-albedo soil puts this project in the 15-25% gain range, well above either of the two projects above.
• Dominant loss factors: at this scale, the ILR/clipping decision has the largest single economic impact of any variable in this piece. NREL's own guidance on bifacial-adjusted ILR applies directly: the optimal ratio here should run lower than a monofacial utility-scale default of ~1.3, specifically because the bifacial rear-side gain is large enough to meaningfully shift the DC output curve the inverter has to handle.
• Decision framework: this project lives or dies on the interaction between bifacial gain and ILR sizing, not on nameplate STC efficiency at all. Two developers using the identical panel model can produce a materially different LCOE purely based on how carefully this interaction was modeled during design, which is the entire argument for running full production simulation rather than a spec-sheet comparison.

What EPCs Should Do Now

Common Mistakes to Avoid

• Don't compare two panels on STC efficiency alone without checking their temperature coefficients, especially for any project in a hot-climate market.
• Don't quote a bifacial premium on a project where the roof or ground surface won't support meaningful rear-side gain.
• Don't resize a panel spec upward for efficiency without revisiting the inverter's DC:AC ratio.
• Don't treat degradation as a footnote; a 0.5%/year difference compounds into a real gap in delivered energy by year 15-20.
• Don't apply a generic national-average system loss percentage to a specific site when actual shading and orientation data is available.

Where Reslink Fits for EPCs Making Panel-Efficiency Decisions

Everything above describes a gap between a spec-sheet number and what a specific site will actually deliver. That gap is exactly what shadow and thermal modeling exists to close, and it's exactly where a generic efficiency comparison runs out of usefulness: it can't account for a specific roof's shading pattern, a specific site's mounting height, or how a specific panel choice interacts with a specific inverter's DC:AC ratio.

Reslink's 3D design runs an 8,760-hour simulation, every hour of every day, for the specific site, modeling real shading from parapets, adjacent structures, and seasonal sun angles rather than applying a flat regional loss assumption. When a panel swap changes the array's effective DC output (a higher-efficiency module, or a move to bifacial), string and inverter configuration is auto-assigned from the updated layout, the direct, practical answer to the ILR-drift problem in Section 6: inverter sizing gets re-checked against the actual design every time the panel choice changes, not left as a stale assumption from an earlier version of the layout.

This runs identically on desktop and mobile, so a panel-technology decision, bifacial versus monofacial, one temperature coefficient versus another, can be modeled and compared on-site, against the specific roof or ground conditions in front of the sales rep, rather than deferred to a follow-up design review.

See how the same shadow and thermal modeling runs on your next project's actual roof, not a generic loss assumption → Book a demo

Frequently Asked Questions

Q1. Is a higher efficiency rating always worth the extra cost?

Not universally. It depends on whether roof or ground space is the binding constraint. On a space-constrained residential roof, higher efficiency means fewer panels for the same target output, which can reduce racking, wiring, and labor costs enough to offset the panel premium. On a site with abundant space, a lower-efficiency, lower-cost panel per watt can produce a better overall system cost, provided enough area exists to reach the target capacity. The decision is project-specific, not a fixed rule.

Q2. Why do two panels with the same STC efficiency perform differently in the same location?

Most commonly, temperature coefficient and degradation rate. Two panels can carry an identical 21.5% STC rating with meaningfully different temperature coefficients (say, -0.40%/°C versus -0.28%/°C), which translates directly into different real-world output on any hot day, and compounds across the system's lifetime.

Q3. Does cleaning solar panels actually improve output enough to matter?

Soiling losses are typically modeled around 2% in NREL's default assumptions, higher in dry, high-traffic, low-rainfall regions where dust accumulates without natural rain washing it off. It's a real, measurable factor, but a smaller one than temperature or degradation for most US climates. Regular rainfall regions see less benefit from manual cleaning than arid regions do.

Q4. Is bifacial worth it for a standard residential install?

Usually not on a dark, flush-mounted asphalt shingle roof, where rear-side gain typically runs 2-5%, often not enough to clear the 3-8% cost premium bifacial modules commonly carry over monofacial equivalents. It becomes a stronger case specifically on elevated mounting over a reflective surface: white membrane commercial roofs, ground-mounts, or carports.

Q5. How does inverter clipping interact with panel choice, in plain terms?

An inverter has a maximum AC output it can produce, regardless of how much DC power the array sends it. Most systems are deliberately designed with more DC capacity than the inverter's AC rating (this is normal and usually the right economic call), but if a panel swap increases the array's effective DC output significantly, either through higher wattage or bifacial gain, without the inverter sizing being re-checked, some of that additional output gets "clipped" away rather than delivered. It's not wasted dramatically at low ILRs, but it climbs quickly at higher ones.

Sources

• National Renewable Energy Laboratory (NREL), PVWatts Calculator and PVWatts Version 5 Technical Manual: system loss model and default loss factor breakdown (soiling, shading, mismatch, wiring, connections, LID, nameplate tolerance, availability)
• Sandia National Laboratories, PV Performance Modeling Collaborative (PVPMC): Sandia Array Performance Model and Sandia Cell/Module Temperature Models, temperature-corrected output methodology
• Jordan, D.C. and Kurtz, S.R., "Photovoltaic Degradation Rates, An Analytical Review," Progress in Photovoltaics: Research and Applications (NREL): median field-measured degradation rate across nearly 2,000 systems
• Jordan, D.C., Kurtz, S.R., and VanSant, K., "Compendium of Photovoltaic Degradation Rates," Progress in Photovoltaics (NREL): updated degradation dataset across 11,000+ measurements, technology-specific findings
• Denholm, P. et al., National Renewable Energy Laboratory: inverter loading ratio (ILR) benchmarking and utility-scale sizing research
• U.S. Energy Information Administration (EIA): observed inverter loading ratio ranges for installed US systems