…and the Rise of Some Truly Terrible Replacements.

For years, Canadian homeowners enjoyed access to something that sounded almost too good to be true… and shockingly wasn’t.
We’re talking about the Canada Greener Homes Loan (CGHL) — the unicorn of financing:

✅ Up to $40,000
✅ 0% interest
✅ Pay back over 10 years
✅ No admin fees
✅ No hidden “surprises”
✅ No penalties for early repayment

A real, honest-to-goodness, government-backed, interest-free loan. Not a financial product wrapped in sparkly marketing. Not a “promo rate gimmick.” Just a genuinely helpful tool.

Alas… that era has ended.
The CGHL portal is now closed to new applications.
Pour one out for the real MVP of solar financing. 🍺

And, as expected, the moment the true-zero-interest train left the station, many solar companies started scrambling.
Some did so ethically…
Others did what you’d expect when the candy bowl disappears — they got desperate.

And now, the financing tactics we’re seeing?
Well… they deserve a spotlight. And maybe a warning siren. 🚨

Let’s talk about it.

🚨 What We’re Seeing — And Why We’re Concerned
❌ “Don’t pay until 2027!”

Sounds great, right?
It’s November 2025. Who doesn’t love the idea of skipping payments for a year?

But here’s the hidden plot twist:
We looked into one of these offers through FinanceIt. Here’s what’s actually happening:

• The contractor pays 6.99% of the loan amount up front to “buy” the 1-year no-payment promo.
• After that year? The homeowner is stuck with 13.99% interest for up to 19 more years.
• And that 6.99% cost? It’s usually buried into the system price… and never mentioned out loud.

So in real terms, the customer:

• Pays more for the solar system (to cover the 6.99%)
• Then pays 13.99% interest for nearly two decades

That “don’t pay for a year” teaser turns into:

The world’s most expensive nap.

This is how you buy a solar system…
and pay for it twice.

Who wins?
✅ FinanceIt
✅ Any installer willing to play along

Who loses?
❌ Homeowners

This is the opposite of what made CGHL special.
This is marketing glitter at payday-loan prices.

And unfortunately, it gets worse…

🚨 Even Worse: Fake “0%” Loans With $10,000+ in Fees

We’re also seeing a separate in-house financing model floating around Nova Scotia. This one has been here for a while folks.
It loudly advertises:

“0% FINANCING — 20 YEARS!”

Sounds like the CGHL reincarnated, right?
Not so fast…

Here’s how it works:

• 20-year repayment term
• A lien on your home
• “Admin fees” over the term that often exceed $10,000+

That’s right:
They didn’t remove the borrowing cost — they just renamed it.

If you pay $10,000 to borrow money, it doesn’t matter whether you call it:

• “Interest”
• “Admin fees”
• “Unicorn glitter handling charges”

It’s still money you’re losing.

At least a traditional bank tells you the interest rate.
Here, the true cost is camouflaged.

This structure can:

• Reduce home-sale flexibility
• Interfere with refinancing
• Lock you into long-term debt

Frankly, we consider this model more misleading than high-interest loans — because it hides the cost behind friendlier words.

“0% interest” should mean…
0% interest.
Not “just kidding, here’s a $10,000 service fee and a lien.”

Big yikes. 🙃

We get it — business is business.
But business must still be ethical.

Anyone who understands solar economics knows that too much interest destroys the financial value of a solar system.
Financing should help homeowners — not sabotage their ROI.

✅ There’s Good News: Municipal PACE-Style Loans Still Exist

Before you lose hope — there are still responsible financing pathways in Nova Scotia.
The unsung hero: Municipal PACE programs.

Programs like:

Halifax Solar City

Various Clean Energy Financing programs

SwitchPACE

These have been around for years — long before the CGHL boom — and they’re run by municipalities, not private loan sharks.

They offer benefits like:
✅ Fair interest rates
✅ Good term lengths
✅ Added to your property tax bill
✅ Transferable on sale
✅ No gimmicks
✅ Transparent pricing

And best of all:
They don’t rely on “buy now, panic later” marketing tricks.

You can browse every NS municipal program on our financing page:

👉 https://wattsupsolar.ca/solar-financing

We recommend them.
Happily.
Enthusiastically.

In the solar industry, we don’t sell short-term products.

We sell 25-year relationships.

Panels carry 30-year warranties. Microinverters carry 25-year warranties. Production guarantees stretch three decades. When a homeowner installs solar they are entering into a long-term partnership with the company that designs, installs, and services their system.

That kind of relationship must be built on one thing:

Trust.

Unfortunately, after 12 years in the Nova Scotia solar industry running Watts Up Solar, we are seeing more deception than ever before.

Experience You Can’t Fabricate

Recently, a company registered with the Nova Scotia Joint Registry of Stocks in November of last year (just five months ago) launched a website (domain also registered five months ago) claiming:

• Nearly 1,000 solar installations
• Installation of a well-known 1 MW utility-scale solar project

The problem?

The company did not install that 1 MW project. The photo displayed beside the claim is not even the correct solar field, and the company did not exist when the project was built.

Let’s be clear: a new company is not automatically a bad company. Every reputable firm started somewhere.

But claiming large-scale experience that did not occur is not “marketing.”

It’s deception.

Installation Claims vs. Reality

Installing thousands solar systems is not something that happens in a few months. It takes years of consistent operations, permitting, inspections, scheduling, and service work to reach that volume. Many established and reputatble solar companies in Nova Scotia who have been doing good work for nearly a decade have not reached a fleet of 1,000 systems.

Solar installations involve:

• Engineering and system design
• Utility interconnection approvals
• Equipment procurement
• On-site installation crews
• Electrical inspections
• Ongoing monitoring and service

A company incorporated five months ago did not install hundreds of systems. That level of volume requires infrastructure, and comes naturally with a documented track record.

Fake Reviews & “Review Farming” in the Solar Industry

Online reviews matter. They help homeowners choose a solar company with confidence. But not all reviews are created equally.

In recent years, we’ve seen a growing tactic in the solar industry known as “review farming.”

This happens when companies request 5-star reviews from people who never actually purchased a solar system. These reviews often come from individuals who simply had a phone consultation or received a quote.

They typically read something like:

“Chris was very informative and professional.”

While that may reflect a positive conversation, it is not the same as a review from a homeowner who invested $30,000, completed an installation, and has lived with the system for months or years.

A consultation is not an installation. A phone call is not a 25-year customer relationship.

An Even Bigger Red Flag: Fake Accounts

More concerning is the use of entirely fabricated Google accounts to generate artificial 5-star reviews.

Patterns consumers should watch for:

• Reviewers who have only reviewed one or two businesses
• Groups of accounts reviewing the exact same unrelated companies
• Generic wording repeated across multiple reviews
• A high volume of reviews for a very new company

Authentic solar reviews should reflect real installations, real timelines, and real customer experiences.

At Watts Up Solar, every one of our 260+ five-star Google reviews corresponds to a completed solar installation. Every review is tied to a real homeowner with a verifiable system.

That’s what transparency looks like.

Why This Matters: A Recent Nova Scotia Example

If this feels dramatic, it shouldn’t.

Just look at what happened with Sun Kissed Energy in Nova Scotia.

CBC reported how homeowners were left out tens of thousands of dollars when the company shut down. Suppliers were owed hundreds of thousands for equipment. Customers were left without systems, without recourse, and without refunds.

When companies operate without transparency and financial stability, real people get hurt.

Solar projects often involve $25,000–$40,000 investments. This is not a small purchase. It is not something homeowners can afford to “roll the dice” on.

How Consumers Can Protect Themselves

Now more than ever, homeowners should:

1. Check the Joint Registry of Stocks – When was the company incorporated?
2. Verify claimed projects – Ask for references, and proof.
3. Request real customer contacts – Not testimonials. Real people.
4. Examine review profiles – Do they look organic?
5. Ask how many systems were installed in the past 12 months.
6. Ask who services the system 10 years from now.

A reputable company will welcome these questions. A deceptive one will deflect.

New Companies Deserve Opportunity, Not a Free Pass to Mislead

This is not an argument that new companies should never compete. Healthy competition improves the industry.

But “fake it until you make it” has no place in a sector built on 25-year warranties and long-term service commitments.

If a company begins by exaggerating or fabricating experience, it has already violated the most important part of the transaction:

Customer trust.

The Solar Industry Needs Transparency

The future of solar in Nova Scotia depends on integrity. It depends on companies being honest about their experience, their capabilities, and their track record.

Trust is not a marketing tactic. It is the product.

And once it’s lost, the entire industry pays the price.

We Talked About Financing Before… And Something Funny Happened

Last year we published:

“The End of True 0% Solar Loans in Canada (…and the Rise of Some Truly Terrible Replacements)”

In that piece, we broke down:

• The end of the Canada Greener Homes Loan
• The rise of “don’t pay until 2027” teaser financing
• The 6.99% dealer fees buried into system pricing
• The 13.99% interest that follows
• The 20-year “0%” loans with $10,000+ in admin fees and liens

We just explained how the math works. Because math doesn’t care about marketing.

Shortly after publishing that article, our FinanceIt account (which we’ve had since 2016 and never used once) stopped working.

No notice.
No conversation.
No explanation.

We’ve never submitted a single loan through it. Not one.

When transparency makes certain financial products harder to sell, reactions can be… revealing. We’re not mad about it. If anything, it reinforces the entire point:

The solar industry works best when financing is clear, experience is verifiable, and companies don’t need smoke machines to close deals.

If speaking plainly about math means losing access to gimmicks? We’ll survive.

Solar is not a discretionary purchase like a vacation. It is not a consumable expense. Solar is an investment asset designed to reduce long-term utility costs and generate financial return over decades. Paying a lot of interest on some products can be justified. A vehicle, for example, provides immediate mobility and utility regardless of financing cost.

Solar is different. Its value proposition is rooted in savings.

When financing reaches 13–14% interest over long amortization periods, the numbers change dramatically. You get one solar system and pay for it two and a half times.

That is not financial optimization. That is erosion of return.

Municipal PACE-style programs in Nova Scotia (including Halifax Solar City and other clean energy financing programs) provide:

• Transparent terms
• Reasonable interest rates
• Transferability upon sale
• Integration into property tax structures
• No artificial “dealer fee” inflation

These programs were available before the Greener Homes Loan, and they remain available now.

Integrity Is the Real Competitive Advantage

Solar does not need exaggeration to sell. It does not require inflated installation counts or borrowed credibility. It does not need 14% interest loans disguised as opportunity. Solar has proven itself, year after year, as one of the most reliable long-term investments available to homeowners.

When installed correctly and financed responsibly a solar system delivers measurable financial return over decades. That is more than enough.

What is disappointing is not competition. Competition is healthy. What is disappointing is watching parts of the industry drift toward short-term tactics in a sector built on 25-year relationships.

But our optimism outweighs our frustration.

Spring is arriving in Nova Scotia 🌷. The days are lengthening. Production numbers are climbing. Homeowners are thinking about the year ahead. We are entering another strong season with exciting projects underway and new technologies coming to market.

The future of solar remains bright because the fundamentals are solid.

Sunlight is predictable.
The math works.
And integrity still matters.

We look forward to another stellar year serving homeowners who value transparency, long-term thinking, and doing things the right way.

A Physically-Based Solar Irradiance and Energy Simulation Framework

Accurate prediction of photovoltaic (PV) system performance requires a detailed understanding of how solar radiation interacts with both the module surface and the surrounding environment. Modern PV installations operate under a wide range of conditions, where geometry, atmospheric effects, and local obstructions all influence energy production. As system complexity increases, particularly with the adoption of bifacial modules, traditional modeling approaches can struggle to capture the full range of physical interactions that drive performance.

Conventional PV performance tools, such as PVWatts and similar industry-standard models, rely on simplified assumptions to estimate energy production. These assumptions typically include uniform ground reflectance, idealized array layouts, and analytical approximations of irradiance components. While effective for large-scale estimation and benchmarking, such approaches do not fully capture the spatial and directional variability of irradiance present in real-world environments, particularly in the presence of shading, complex terrain, or non-uniform surfaces.

A comprehensive PV simulation framework must resolve several interacting physical processes, including solar position, anisotropic sky diffuse behavior, angle-dependent optical transmission, and geometry-dependent reflection. In bifacial systems, these challenges are further amplified by the need to accurately model rear-side irradiance, which is highly sensitive to local visibility conditions and the spatial distribution of incident radiation on surrounding surfaces.

This work presents a physically based solar irradiance and energy model designed to bridge the gap between simplified industry tools and higher-fidelity simulation approaches. The framework decomposes incident solar radiation into direct, diffuse, and ground-reflected components, evaluated independently for each module surface. Diffuse irradiance is modeled using an anisotropic sky formulation derived from the Perez model, while ground-reflected contributions are treated using a hybrid approach that combines analytical view factors with geometry-aware visibility estimation.

To maintain computational efficiency while preserving physical accuracy, the model employs spatially resolved sampling of solar irradiance and directional visibility. This enables accurate representation of occlusion effects and non-uniform reflection without requiring full radiative transfer simulation. Under equivalent assumptions, the formulation reproduces standard analytical models, ensuring consistency with established tools such as PVWatts. In more complex environments, however, it enables higher-fidelity evaluation by explicitly accounting for scene-dependent effects.

The resulting system provides a complete energy conversion pipeline, linking irradiance to DC and AC output through optical, thermal, and electrical models. These include angle-of-incidence-dependent transmission, temperature-dependent module performance, and inverter conversion with clipping and loss mechanisms. Bifacial performance is naturally supported within this framework through explicit modeling of front and rear irradiance contributions.

Validation is performed through comparison with PVWatts under equivalent assumptions, as well as against published production ranges across varying environmental conditions and system configurations.

Overall, the approach provides a practical balance between physical realism and computational efficiency, enabling accurate and responsive prediction of PV system performance across a wide range of operating conditions while remaining compatible with industry-standard modeling methodologies.

Geometry-Aware Shading and Irradiance Redistribution

A key component of the proposed framework is the explicit treatment of shading as a physically resolved, geometry-dependent process rather than a fixed loss factor.

Unlike conventional models such as PVWatts, which typically apply shading as a uniform percentage reduction, the present model evaluates shading through direct interaction with scene geometry. Occlusion is computed at the irradiance sampling level, allowing each component of incident radiation to be affected independently based on its directional origin.

• Direct irradiance (beam) is reduced through explicit ray-based intersection with scene geometry, producing time-varying shadows consistent with solar position.
• Diffuse sky irradiance is modulated by directional visibility, where obstructions such as vegetation or nearby structures partially occlude the sky dome. This results in a continuous reduction of diffuse contribution rather than binary shadowing.
• Ground-reflected irradiance is influenced both by visibility of the ground and by the spatial distribution of incident radiation on ground surfaces. Shadowing of the ground itself reduces reflected energy, particularly impacting bifacial systems.
• Coupled effects emerge naturally, as energy blocked from one component is not artificially preserved in another. This leads to physically consistent redistribution of irradiance across beam, diffuse, and reflected terms.

This approach enables accurate representation of complex shading scenarios, including partial occlusion, soft shadowing of diffuse components, and secondary effects on ground reflection. As a result, system performance is evaluated under realistic environmental conditions without reliance on empirical shading factors.

Monofacial System Performance (Baseline Behavior)

Conventional monofacial photovoltaic systems generate energy solely from irradiance incident on the front surface of the module. In these systems, energy production is primarily driven by direct beam irradiance and diffuse sky radiation, with ground-reflected contributions playing a secondary role.

As a result, monofacial system performance is largely determined by solar geometry, module orientation, and atmospheric conditions. Analytical models, such as those used in PVWatts, provide accurate estimates under these assumptions by treating the environment as spatially uniform and relying on view-factor-based approximations for diffuse and reflected components.

Within the presented framework, monofacial behavior emerges naturally by excluding rear-side contributions. This provides a direct baseline for validation, ensuring that front-side irradiance modeling and energy conversion remain consistent with established industry methodologies before extending to bifacial scenarios.

Bifacial Systems and Albedo-Driven Performance

Bifacial photovoltaic modules are designed to generate energy from both the front and rear surfaces, enabling additional energy capture from reflected and diffuse irradiance. Unlike conventional monofacial systems, where performance is dominated by direct and sky diffuse irradiance on the front surface, bifacial systems introduce a strong dependence on ground conditions and surrounding geometry.

Rear-side energy production is primarily driven by ground-reflected irradiance, which is a function of surface reflectivity (albedo), module elevation, and the visibility of surrounding surfaces. As a result, bifacial system performance can vary significantly depending on installation design and site characteristics.

In natural conditions, typical ground surfaces such as vegetation or soil exhibit albedo values in the range of 0.15–0.25, corresponding to modest bifacial gains. However, engineered system designs can significantly enhance rear-side contribution through deliberate modification of ground reflectance. Common approaches include the use of high-reflectivity materials such as white gravel, reflective membranes, or specialized coatings, which can increase effective albedo and drive substantially higher energy yields.

These design strategies introduce a nonlinear relationship between albedo and energy production, where incremental increases in reflectivity can result in disproportionately larger gains in bifacial output. As a result, accurate modeling of ground-reflected irradiance and visibility becomes critical for both system optimization and financial forecasting.

Within this framework, bifacial performance is not treated as a fixed multiplier, but rather as an emergent property of the system, driven by the interaction between irradiance distribution, geometry, and material properties. This enables the model to capture both baseline performance under typical conditions and enhanced production scenarios enabled by engineered site design.

Irradiance Model

Accurate prediction of photovoltaic system performance depends on correctly resolving how solar radiation interacts with the module surface and its surrounding environment. This interaction is governed by a combination of solar geometry, atmospheric conditions, and local scene characteristics, all of which influence the distribution of incident energy across the system. As system complexity increases, particularly in the presence of non-uniform terrain, shading, or advanced module technologies, simplified assumptions can lead to meaningful deviations in predicted energy yield.

To address this, the model decomposes incident solar radiation into three physically distinct components: direct sunlight, diffuse sky radiation, and ground-reflected irradiance. Each component is modeled independently to capture its unique dependence on orientation, atmospheric conditions, and surrounding geometry.

Direct irradiance represents the portion of solar energy arriving from the sun and is governed primarily by the angle of incidence relative to the module surface. Diffuse irradiance arises from atmospheric scattering and is distributed across the sky dome, requiring an anisotropic representation to capture circumsolar and horizon effects. Ground-reflected irradiance depends on both surface reflectivity and the fraction of visible ground surrounding the array, introducing a strong dependence on system geometry and local environmental conditions.

In bifacial systems, these interactions are further extended through the inclusion of rear-side irradiance, which introduces additional sensitivity to ground conditions, mounting height, and occlusion. As a result, accurate modeling of ground reflection and directional visibility becomes increasingly important for capturing system performance.

The model combines established analytical formulations with geometry-aware visibility constraints to balance accuracy and computational efficiency. Front-side ground reflection is computed using a bounded view-factor formulation constrained by directional sky visibility, ensuring physically consistent results in partially occluded environments. Geometry-dependent effects, including shading and reflection, are further resolved using spatially aware sampling, enabling the model to capture non-uniform irradiance distributions that are not represented in conventional approaches.

Under equivalent assumptions, this formulation reproduces standard view-factor-based models, maintaining consistency with industry tools such as PVWatts. In more complex environments, however, it provides a higher-fidelity representation of system performance by explicitly accounting for scene-dependent effects, including those that are particularly impactful in bifacial installations.

The resulting plane-of-array irradiance is expressed as:

GPOA= GB+Gd+Gr

where Gbis the beam (direct) component, Gd​ is the diffuse sky component, and Gr​ is the ground-reflected component. These terms are evaluated independently for the front and rear surfaces and combined using the module’s bifaciality factor.

Direct Component

The direct component is computed from direct normal irradiance (DNI) projected onto the module surface:

Gb =DNImax(0,i)

where i is the angle of incidence between the sun vector and the module normal. This term inherently captures orientation and tilt dependence and is constrained to non-negative contributions.

Diffuse Sky Component

Diffuse irradiance is modeled using an anisotropic sky formulation derived from the Perez sky model, which accounts for circumsolar brightening and horizon effects. The general form is:

Gd = DHI [F1iz+F2 + (1-F1)1 + 2]

where:

• DHI is diffuse horizontal irradiance
• i is solar zenith angle
• is surface tilt
• F1 F2​ are empirically derived anisotropy coefficients

This formulation provides a directional weighting of sky irradiance rather than assuming isotropic distribution.

Ground-Reflected Component (Front Surface)

Ground-reflected irradiance for the front surface is modeled as a Lambertian reflection of incident irradiance from the surrounding ground, scaled by the effective ground albedo and constrained by both geometric view factors and local visibility.

The front-front side ground-reflected component is expressed as:

Gr pGHIVground

Where:

Vground = min((1 - )2 , 1-Vsky)

And:

• is ground albedo
• GHI is global horizontal irradiance
• is surface tilt
• VSky is the sky visibility factor derived from directional sampling
• Vground represents the effective ground view factor

This formulation combines the standard view-factor-based approximation:

1 - 2

Ground-Reflected Component (Rear Surface)

Rear ground-reflected irradiance is modeled as a Lambertian reflection of incident irradiance from surrounding surfaces, scaled by the effective ground albedo and constrained by geometry-dependent visibility. Rather than relying on a purely view-factor-based approximation, the model evaluates reflected contributions using spatially resolved proxy sampling of incident beam and diffuse irradiance.

The ground-reflected component is therefore expressed as:

Gr ρ(Gbground+Gdground)

where:

• ρ is the ground albedo
• Gbground​ is the beam irradiance incident on the ground and reflected toward the module
• Gdground is the diffuse irradiance incident on the ground and reflected toward the module

These terms inherently account for occlusion, array geometry, and directional visibility effects.

Under simplified conditions, this formulation reduces to the standard view-factor-based approximation:

Gr ρGHI1-cosβ2

ensuring consistency with conventional models while enabling higher-fidelity evaluation in complex scenes.

Bifacial Combination

Rear-side irradiance is scaled by the module bifaciality factor kbk_bkb​ and combined with the front-side contribution:

Gtotal = Gfront + kb Grear

Typical values of kb​ range from 0.65 to 0.85 depending on module design.

System Conversion

The modeled irradiance is first adjusted for optical transmission through the module cover using an incidence-angle modifier (IAM). Beam, sky-diffuse, and ground-reflected components are evaluated separately to account for their different effective angles of incidence:

Gefffront = IAMb GB + IAMd Gd + IAMr Gr

For bifacial modules, rear-side effective irradiance is computed analogously for the rear-facing surface and scaled by the bifaciality factor:

Gtotal = Gefffront + kb Geffrear

The effective irradiance is then converted to DC output using a temperature-dependent module performance model:

PDC = PSTC (Gtotal1000) [1 +γ (Tcell - 25) ] (1 - LDC)

Cell temperature is estimated from irradiance, ambient temperature, wind speed, mounting configuration and module efficiency:

Tcell = fnoct (Gtotal , Tambient, vwind, NOCT, module)

Finally, DC output is converted to AC output using an inverter performance curve with clipping and transformer losses:

PAC = min(finv(PDC), PAC, rated) - Lxfmr

Overall, this formulation represents a complete system-level energy conversion model, linking incident irradiance to AC output through optical, thermal, and electrical processes. The approach preserves consistency with established PV performance models while incorporating geometry-dependent bifacial effect, allowing accurate representation of real-world system behaviour across a wide range of operating conditions.

Model Characteristics

This formulation captures the primary physical drivers of bifacial performance:

• Dependence on solar geometry (tilt, azimuth, incidence angle)
• Directional diffuse sky behavior
• Ground reflectance and its nonlinear contribution to rear irradiance
• Separation of front and rear irradiance pathways

The model is designed to maintain consistency with established industry tools under equivalent assumptions while enabling extension to more complex geometries and environmental conditions.

GPU-Accelerated Computation

Accurate evaluation of bifacial irradiance requires resolving directional visibility, spatial occlusion, and time-dependent solar geometry across a large number of samples. Traditional CPU-based approaches often rely on simplified analytical approximations to remain computationally tractable, limiting their ability to capture scene-dependent effects.

To address this, the model leverages GPU-accelerated computation to evaluate irradiance and visibility at scale, enabling physically based modeling without compromising performance.

The simulation is formulated as a data-parallel problem, where irradiance contributions are evaluated independently across spatial samples (module surfaces) and temporal samples (solar positions). By mapping these evaluations to GPU execution, the system efficiently processes large datasets, including full-year (8760-hour) simulations and high-resolution panel-level calculations.

Directional visibility and irradiance sampling are performed using a ray-based approach, allowing the model to resolve occlusion and shading effects directly from scene geometry. This enables accurate estimation of both sky access and ground-reflected contributions, which are critical for bifacial performance. The approach avoids reliance on precomputed shading factors or simplified horizon models, instead deriving visibility dynamically from the environment.

To maintain efficiency, the system combines analytical models with targeted sampling. Direct and diffuse irradiance components are evaluated using established formulations, while geometry-dependent terms—such as ground reflection and visibility—are resolved through GPU-based sampling. This hybrid strategy ensures that computational effort is focused on the aspects of the problem where spatial detail is most impactful.

The use of GPU acceleration enables near real-time evaluation of complex systems, including thousands of modules across full annual simulations. This allows for rapid iteration, sensitivity analysis, and visualization of system behavior under varying conditions, while maintaining physical consistency with established irradiance and performance models.

Overall, this approach provides a scalable computational framework that supports high-fidelity bifacial modeling, bridging the gap between analytical efficiency and physically based simulation.

Validation and Calibration

The modeling framework is rigorously benchmarked, calibrated, and validated against established industry and research standards, including NREL PVWatts and SAM-derived formulations. Core irradiance components—such as plane-of-array (POA) calculations and temperature-dependent performance—are systematically cross-validated under controlled conditions to ensure consistency with widely accepted methodologies.

The model is explicitly constructed to reproduce standard analytical formulations under equivalent assumptions, maintaining direct comparability with conventional tools. At the same time, it extends beyond these approaches through physically based treatment of ground-reflected irradiance and geometry-dependent visibility, enabling accurate representation of effects that are not captured by traditional models.

This approach ensures that results are both bankable and high-fidelity: consistent with industry benchmarks while providing a more complete representation of real-world system behavior. The underlying GPU-accelerated architecture further enables these calculations to be performed with high responsiveness, supporting rapid iteration and interactive system design without compromising physical accuracy.

Baseline (Front-Side) Production Validation Summery  

To maintain consistency with established models while extending physical realism, the front-side irradiance formulation is rigorously benchmarked and validated against industry-standard tools, including PVWatts and SAM-derived formulations. Under equivalent assumptions, the model reproduces baseline behavior, providing direct comparability with conventional approaches.

Building on this validated foundation, the formulation incorporates several geometry-aware refinements. Ground-reflected irradiance is modeled using a simplified Lambertian formulation, scaled by surface albedo and attenuated by an effective ground visibility factor derived from directional sky visibility terms. This ensures that ground contributions are bounded by actual scene visibility rather than assuming an unobstructed hemisphere, while remaining computationally efficient.

In addition, the model applies an optical transmission adjustment to plane-of-array irradiance through a physically based incidence-angle modifier (IAM), resulting in a transmitted plane-of-array (TPOA) value used for energy conversion. This accounts for angle-dependent reflection and absorption losses within the module cover, which are typically simplified or omitted in baseline analytical models.

Diffuse irradiance is evaluated using a directional sky formulation that explicitly accounts for partial occlusion of the sky dome. Unlike conventional tools such as PVWatts and SAM, which assume full sky visibility for diffuse irradiance regardless of geometry, this model resolves sky access based on scene-dependent visibility. As a result, portions of diffuse irradiance may be redistributed or reduced when obstructions or ground interactions limit exposure to the sky.

These effects introduce small but physically consistent deviations from PVWatts under non-ideal conditions. In particular, reductions in effective sky diffuse irradiance and the inclusion of optical transmission losses tend to produce slightly lower front-side energy estimates, while improved modeling of ground interaction redistributes irradiance in a manner consistent with real-world installations. When these effects are disabled or reduced to PVWatts-equivalent assumptions, the model converges closely to baseline outputs, confirming that observed differences are attributable to intentional physical modeling rather than numerical bias.

Deviations from PVWatts (Front-Side Modeling)

Deviations from PVWatts on the front-side occur when physical effects not represented in its analytical formulation are enabled. To establish a consistent baseline, two reference cases are considered:

• PVWatts (No Shading): Represents the idealized analytical assumption of full-sky and ground exposure
• PVWatts (Adjusted, −3% Shading): Represents a more typical real-world installation baseline

The model is capable of reproducing both reference conditions closely, demonstrating alignment with industry-standard energy estimation under equivalent assumptions.

Beyond this baseline, the model explicitly incorporates:

• Optical transmission losses (TPOA)Angle-dependent reflection and absorption through the module cover reduce effective irradiance relative to raw plane-of-array assumptions.
• Geometry-aware sky visibilityDiffuse sky irradiance is reduced or redistributed when portions of the sky dome are occluded by terrain or nearby structures.
• Ground visibility-constrained reflectionGround-reflected irradiance is limited to physically visible surfaces rather than assuming an unobstructed hemispherical ground contribution.
• Directional (anisotropic) diffuse response under occlusionThe Perez sky model is modulated by visibility, altering the relative contribution of circumsolar and horizon components.
• Coupled ground–sky interaction Energy intercepted by the ground is not simultaneously available to the sky term, producing a physically consistent redistribution of irradiance.
• Scene-dependent irradiance redistributionEnergy shifts dynamically between beam, diffuse, and reflected components based on local geometry rather than remaining fixed under uniform assumptions.

Table 1. Front-Side Energy Production as a Function of Model Refinement

For unshaded rows, deviation is computed relative to PVWatts (No Shading). For shaded-equivalent rows, deviation is computed relative to PVWatts (Adjusted −3% Shading).

Interpretation of Results

The progressive deviation from PVWatts is expected and physically meaningful.

The PVWatts-equivalent configuration reproduces the unshaded analytical baseline within +0.92%, confirming that the core irradiance and energy conversion pipeline is consistent with established models. When a typical shading loss (~3%) is introduced, the adjusted PVWatts baseline aligns closely with the fully constrained model output, further validating real-world applicability.

As additional physical effects are enabled, total energy decreases because the model no longer assumes:

• Full hemispherical sky exposure
• Unrestricted ground reflection
• Ideal optical transmission

Instead, it resolves irradiance under actual geometric constraints, resulting in a more conservative and physically grounded estimate.

This behavior does not represent model drift. Rather, it reflects a transition from an idealized analytical model to a scene-aware physical model.

Importantly, the model remains within a narrow deviation range relative to PVWatts while incorporating effects that more closely resemble System Advisor Model-style physical modeling and real installation conditions.

Model Convergence Under Analytical Assumptions

To establish a consistent validation baseline, the proposed model is first evaluated under the same simplifying assumptions used by PVWatts. These include full sky visibility, unobstructed ground reflection, and the absence of shading losses.

While the underlying formulations differ—particularly in the treatment of diffuse irradiance, angular response, and spatial sampling—the models converge under these conditions because they are driven by the same fundamental irradiance inputs (DNI/DHI) and energy conversion relationships.

In this configuration, the present model effectively reduces to an analytical-equivalent representation of plane-of-array irradiance, where directional and spatial effects collapse to uniform assumptions consistent with PVWatts. As a result, agreement between the two models is expected and serves as a validation of the core irradiance and energy conversion pipeline.

Deviations observed in subsequent configurations arise not from model inconsistency, but from the progressive relaxation of these assumptions. As physically based effects such as optical transmission, geometry-aware visibility, and explicit shading are introduced, the model transitions from an analytical approximation toward a more complete representation of real-world operating conditions.

Table 2. PVWatts Equivalence Across Orientation (Unshaded)

Max deviation: 0.92%Mean absolute deviation: 0.41%

Across all tested tilt and azimuth configurations, the model demonstrates strong agreement with PVWatts under unshaded conditions. Deviations remain within ±1%, with a mean absolute deviation of approximately 0.41%, indicating a high degree of consistency with established analytical results.

No systematic directional bias is observed across orientations, with comparable accuracy maintained for south-facing, east/west-facing, and off-axis configurations. Minor variation in deviation is expected and can be attributed to differences in diffuse irradiance treatment and angular response modeling, particularly under varying tilt conditions.

Overall, the results confirm that the model reliably reproduces PVWatts behavior across a representative range of system orientations, providing a consistent baseline for further physically based refinement.

Physically Based Shading vs. Uniform Loss Assumptions

In conventional tools such as PVWatts, shading is typically represented as a single aggregate loss factor applied uniformly to total system output. While this approach provides a convenient approximation, it does not distinguish between different irradiance components or account for the spatial and directional nature of shading.

In contrast, the present model evaluates shading through explicit interaction with scene geometry. Occlusion is resolved at the irradiance sampling level, allowing each component of incident radiation to be affected independently:

• Direct irradiance (beam) is reduced through time-dependent geometric shadowing.
• Diffuse sky irradiance is partially attenuated based on directional visibility of the sky dome.
• Ground-reflected irradiance is influenced both by visibility of the ground and by shadowing of the ground itself.

This component-wise treatment enables physically consistent redistribution of energy rather than uniform reduction. As a result, portions of diffuse or reflected irradiance may remain available even when direct irradiance is obstructed, leading to small but systematic differences relative to uniform-loss models.

The observed deviation from PVWatts under shaded conditions, including the slight positive bias, is therefore expected and reflects a more realistic representation of irradiance behavior in complex environments.

Table 3. PVWatts Equivalence Across Orientation (Shaded)

Max deviation: 1.63%Mean absolute deviation: 0.81%

Across all tested tilt and azimuth configurations, the model demonstrates strong agreement with PVWatts when compared against a shading-adjusted baseline. Deviations remain within approximately ±2%, with a mean absolute deviation of ~0.90%, indicating close alignment under representative shaded conditions.

A small positive bias is observed across most configurations, which is expected. Unlike PVWatts, which applies shading as a uniform loss factor, the present model resolves shading explicitly through scene geometry. This allows partial and directional irradiance contributions to be preserved where physically appropriate, rather than uniformly reduced.

No strong directional dependence is observed, with consistent behavior across south-facing, east/west-facing, and off-axis configurations. Minor variation arises from the interaction between shading geometry and the directional distribution of irradiance components.

Overall, the results confirm that the model remains well-aligned with PVWatts under shaded assumptions while providing a more physically grounded representation of shading effects.

Bifacial Production Validation Summary

We validated the model against PVWatts under PVWatts-equivalent conditions (uniform row spacing, infinite row assumptions, constant ground plane, and no localized geometry effects). Under these constraints, both front-side and bifacial rear-side production closely match PVWatts outputs, confirming alignment in irradiance modeling, system losses, and baseline energy conversion.

To validate bifacial response beyond simplified assumptions, we performed controlled parameter sweeps across a range of ground albedo values and compared the resulting production gains against published experimental and simulation data (Soltec white papers, ScienceDirect studies, MDPI publications). The model reproduces the expected nonlinear relationship between albedo and bifacial gain, with results falling within accepted industry ranges.

Deviations from PVWatts occur when scene geometry departs from its underlying assumptions. Unlike PVWatts, this model explicitly resolves:

• Finite row lengths and edge effects
• Panel elevation above ground
• Row-to-row self-shading and ground occlusion
• Local terrain and obstruction shading
• Directional sky visibility and anisotropic diffuse response

These factors reduce or redistribute rear-side irradiance in realistic installations and are expected to produce differences relative to PVWatts’ idealized infinite-array model.

Overall, bifacial gain increases from approximately 5–10% at albedo ~0.2 (vegetation) to 20–30% at albedo ~0.5 (engineered reflective surfaces), consistent with published literature. Higher gains (25–35%+) are achievable under high-reflectivity conditions such as snow-covered ground.Table 4. Literature-Reported Bifacial Gain as a Function of Surface Albedo (ρ)

Bifacial gain ranges compiled from NREL reports, IEA PVPS Task 13 studies, and peer-reviewed literature, showing strong dependence on surface albedo (ρ) and system geometry.

Bifacial gain increases from ~5–10% at albedo ~0.2 to ~20–30% at albedo ~0.5 under typical ground-mounted conditions, consistent with published experimental and modeling studies (Soltec white papers; ScienceDirect; MDPI).Correlation With Internal Simulation Outputs

To further validate model consistency, we generated internal simulation results across the same albedo range using representative ground-mounted configurations (fixed tilt, typical GCR, and elevated mounting). These results demonstrate strong correlation with published data, with bifacial gain trends aligning in both magnitude and slope.

Minor variations between simulated and published values are attributable to:

• Differences in assumed geometry (height, spacing, tilt)
• Use of spatially-resolved ray-based irradiance vs. simplified view-factor models
• Inclusion of localized shading and occlusion effects

This confirms that the model not only matches established tools under equivalent assumptions, but also generalizes correctly to more realistic installation scenarios.

The following table compares internal simulation outputs against expected industry ranges, demonstrating agreement across the full albedo spectrum

Table 5. Bifacial Gain Validation Across Albedo (Model vs. Literature)

Simulations were conducted for a representative ground-mounted system configuration, with modules elevated approximately 1 meter above ground, south-facing, and tilted at 45°, consistent with typical North American array geometry. ☒ Simulation configuration: South-facing (azimuth 180°) bifacial array with 45° tilt and 1.0 m ground clearance. Module bifaciality was set to 0.7, representative of standard modern bifacial modules.A bifaciality factor of 0.70 was assumed, reflecting standard modern bifacial module performance. Across the evaluated albedo range, the model produces gains that align with published industry and academic expectations. The observed increase in bifacial gain at higher albedo is driven primarily by the model’s physically based treatment of rear-side irradiance, including geometry-aware sky visibility and ground-reflected contributions, rather than reliance on elevated bifaciality assumptions.

Under simplified conditions, the model converges to standard view-factor-based formulations, ensuring consistency with conventional tools while enabling higher-fidelity evaluation in complex geometries.

Conclusion

This work presents a physically based solar irradiance and energy simulation framework designed to bridge the gap between simplified analytical tools and high-fidelity, geometry-aware modeling approaches. By decomposing irradiance into direct, diffuse, and ground-reflected components and evaluating each within a consistent physical framework, the model provides a comprehensive representation of photovoltaic system behavior across a wide range of operating conditions.

Validation against established industry tools, including PVWatts and SAM-derived formulations, demonstrates that the model reproduces baseline performance under equivalent assumptions. Controlled deviations observed under more detailed configurations are both expected and physically justified, arising from the inclusion of optical transmission losses, directional sky visibility, and geometry-constrained ground reflection. These refinements result in a more conservative and realistic estimate of system performance, particularly in non-ideal environments.

For bifacial systems, the framework extends naturally to capture rear-side irradiance as an emergent property of system geometry, surface reflectivity, and directional visibility. Comparison with published literature confirms that the model reproduces expected bifacial gain trends across a wide range of albedo conditions, while also capturing site-specific effects not represented in conventional approaches.

The use of GPU-accelerated computation enables these physically based calculations to be performed efficiently at scale, supporting full-year simulations, panel-level resolution, and near real-time feedback. This allows the model to maintain a practical balance between physical accuracy and computational performance, making it suitable for both detailed engineering analysis and interactive system design.

Overall, the framework provides a robust and extensible foundation for photovoltaic performance modeling, combining alignment with industry-standard methodologies with the ability to resolve complex, real-world effects. This enables more accurate prediction of system behavior, improved sensitivity to installation conditions, and greater confidence in both engineering and financial outcomes.

This approach enables a new class of solar modeling tools that are both physically grounded and interactively responsive, supporting better-informed design decisions in increasingly complex deployment scenarios.

Citations :[1] NREL,Bifacial Photovoltaic Performance Modeling,https://docs.nrel.gov/docs/fy19osti/74090.pdf

[2] ScienceDirect,Bifacial PV performance under varying albedo conditions,https://www.sciencedirect.com/science/article/pii/S2590123025039295

[3] ResearchGate Study,Modeling bifacial PV performance under variable albedo,https://www.researchgate.net/publication/399426430

[4] IEA PVPS,Task 13: Bifacial Photovoltaic Modules and Systems,https://iea-pvps.org/wp-content/uploads/2021/04/IEA-PVPS-T13-14_2021-Bifacial-Photovoltaic-Modules-and-Systems-report.pdf

[5] Natural Power,Bifacial PV and Albedo Impact Study,https://www.naturalpower.com/mediaLibrary/other/english/2395.pdf

[6] MDPI,Optimization of Bifacial PV Systems,https://www.mdpi.com/1996-1073/18/16/4443

[7] Research Study,Spectral Albedo Effects in Bifacial PV Performance,https://www.researchgate.net/publication/357313986