Every solar PV project usually starts with one simple number: expected annual PV yield. At first glance, this number looks clear and precise. In reality, however, it always comes with uncertainty.

How a project team deals with this uncertainty matters a lot. It influences engineering design, investor expectations, and bank financing, often in ways that only become visible late in the project lifecycle.

In this article, we explain why actively reducing uncertainty is more effective than simply reporting it. We also share practical ways to achieve this through smarter PV yield simulations.

Using a simple project example, we show how different approaches to uncertainty can lead to very different returns and create distinct risks for project engineers, investors, and banks. The examples are illustrative and are not intended to replicate real project decision-making in detail, but to demonstrate how uncertainty reduction can unlock optimization opportunities for all key stakeholders.

For project engineers: clarity enables optimization

Engineers rely on yield estimates to make design decisions. When those estimates are robust and uncertainty is well understood, engineers can design with confidence, tailoring the system precisely to site conditions. When uncertainty is poorly understood or ignored, engineers lose that precision.

Design choices become conservative by default. Safety margins grow. Optimization gives way to approximation. Sometimes this means adding more equipment than necessary, or relying on conservative sizing choices in electrical design, layout, and loss assumptions. Other times it means leaving performance on the table. Either way, uncertainty limits engineers’ ability to extract full value from the site.

Design optimizations enabled by lower uncertainty include:

  • Tighter electrical sizing. Using more accurate data allows the project team to avoid adding extra electrical equipment “just in case.” The system is sized more precisely for how the plant will actually operate, reducing unnecessary components.
  • Simplified balance-of-plant design. Better understanding of the system allows small simplifications in supporting infrastructure such as cables, structures, and auxiliary equipment, without affecting performance or reliability.
  • More accurate loss modelling. Instead of assuming conservative, fixed losses, the project uses better data and models to estimate real energy losses. This avoids over-compensating for losses that are unlikely to occur.

For investors: uncertainty reshapes expectations

Investors rarely assess a project based solely on its expected return. Instead, they focus on the range of possible outcomes and the resilience of returns under less favorable conditions. When yield uncertainty is clearly quantified and transparently communicated, investors can evaluate risk more effectively and form realistic expectations about long-term performance.

When uncertainty is ignored or understated, projects may appear attractive initially, but confidence often weakens during due diligence as underlying assumptions are challenged. In such cases, perceived downside risk increases and the investment case becomes harder to defend.

For this reason, investors place significant value on reliable, long-term production data. Robust data quality is required to make defensible base case, supports downside analysis, and facilitates alignment with lender requirements, which are typically assessed using more conservative assumptions.

distribution

Fig. 1. The probability distribution of PV energy yield illustrates how expected yield decreases as the PXX level increases.

For banks: uncertainty becomes credit risk

Banks incorporate uncertainty into lending decisions by assessing a project’s ability to meet its debt obligations under conservative assumptions. When uncertainty is clearly quantified and well understood, financing structures can be designed more efficiently.

In standard project finance practice, base-case cash flows are usually based on P50 energy estimates. As yield uncertainty increases, the difference between P50 and more conservative production scenarios, such as P90, becomes larger.

Lenders address this risk mainly through structural and covenant-based mechanisms. Covenants are contractual requirements that obligate the project to meet defined financial or operating thresholds. Typical measures include:

  • Lower loan amounts compared to project value (loan-to-value limits), so the bank lends less and takes less risk.
  • Minimum cash-flow safety margins (DSCR requirements), meaning the project must generate more cash than is needed to pay its debt.
  • Required reserve accounts, where cash is set aside in advance to cover debt payments and maintenance costs if revenues are lower than expected.
  • Limits on dividend payments, so cash stays in the project until financial conditions are strong enough.
  • Additional guarantees or support from the project sponsor, which provide extra protection if the project underperforms.

These measures are designed to protect lenders while keeping the base-case financial model intact. As a result, yield uncertainty directly affects how much debt a project can raise and the terms under which it is financed.

Advantages of including uncertainty in calculations and reports

When uncertainty is ignored, stakeholders are forced to protect themselves with conservative assumptions. Engineers fall back on generic safety margins, investors struggle to understand downside exposure, and banks respond by limiting loans or tightening terms. 

Tab. 1. Differences between ignoring and reporting uncertainty for key project stakeholders.

 

Ignoring uncertainty

Calculating & reporting uncertainty

Engineers

Rely on generic assumptions and conservative safety margins

Align design decisions more closely with site-specific conditions

Investors

Downside returns are unclear and difficult to evaluate

Return range becomes visible and risks are better understood

Banks

Apply conservative assumptions that limit debt sizing

Structure loans with greater confidence and predictability

Calculating and reporting uncertainty improves transparency, but it does not fully eliminate these frictions, risk is better understood, yet still priced defensively.

venn uncertainty

Fig. 2. Reducing PV yield uncertainty is beneficial for each stakeholder’s objective

 

More accurate PV yield simulations to reduce uncertainty

To reduce uncertainty, it is important to understand the different factors that influence it. In PV yield simulations, uncertainty comes both from the models used by the software and from the external inputs provided by the user.

Many uncertainty-reduction measures can be implemented at low or even zero additional cost, depending on the tools already used in the project. In most cases, the key is choosing the right software and understanding how input data and models are handled internally. Uncertainty assessments must be always driven by validated data.

Practical actions include:

  • Use reliable solar irradiance datasets based on proven models and long-term satellite observations.
  • Use datasets covering long time periods to better evaluate interannual variability caused by natural weather cycles.
  • Avoid relying only on averaged datasets such as Typical Meteorological Years (TMY), which can hide important weather patterns. When possible, simulate full historical time series.
  • Use sub-hourly input data to capture short-term effects such as power peaks, inverter clipping, and thermal behavior, which are often hidden when data is averaged to hourly or longer intervals.
  • Run simulations using advanced models that reflect the real complexity of PV plant design, such as ray-tracing for optical losses.
  • Estimate additional losses using physics-based models instead of fixed assumptions. Replace rules of thumb with models for effects such as soiling, albedo, and temperature.
  • Review and validate manufacturer technical datasheets to ensure that modeled component performance matches what will actually be installed.

Some additional actions require extra investment, such as installing ground-based measurement equipment. While this increases upfront cost, it is usually small compared to total power plant CAPEX. In complex locations or regions with limited validation of satellite data, complementing satellite irradiance data with on-site measurements and locally adapted models can significantly reduce uncertainty and improve confidence in the results.

Accepting current uncertainty vs actively reducing it

A project can either accept uncertainty as it is, or take steps to actively reduce it. When uncertainty is reduced, yield estimates become more reliable. This allows engineers to optimize designs with greater confidence, helps investors improve minimum expected returns, and gives banks more comfort to structure debt efficiently. 

The impact of these two approaches is illustrated below through two project scenarios.

Tab. 2. Differences between accepting and reducing uncertainty for key project stakeholders.

 

Accepting current uncertainty

Actively reducing uncertainty

Engineers

Design remains conservative and less competitive

Design is optimized while minimizing overdesign and underdesign risks

Investors

Investment case is weaker due to a wide gap between expected and conservative returns

Minimum expected returns improve and downside risk is reduced

Banks

Higher risk exposure drives caution in financing decisions

Financing costs decrease and capital structure becomes more efficient

 

Scenario A – “Let’s do nothing”

Let’s illustrate this with a sample project: a company is developing a 10 MW solar PV project. The site looks good, the layout is clean, and the expected production is 1,500 kWh per kWp. On paper, everything works.

The project engineer

The engineer runs the yield simulation using standard inputs and a limited dataset. This includes default assumptions instead of physically modelled losses, averaged datasets such as TMY rather than full time series, and hourly simulations instead of sub-hourly resolution. The result looks reasonable, but uncertainty remains relatively high.

  • Expected production (P50): 15 GWh/year
  • PV yield Uncertainty: ±10%
  • Conservative production (P90): 13.5 GWh/year

The engineer knows the numbers are not wrong, but also knows they are not very tight. So the team decides to move forward as is.

Tab. 3. Uncertainty assumptions used in Sample Scenario A.

Uncertainty component

Value

Notes

Solar irradiance data uncertainty

±7.0%

Standard satellite-based datasets with limited local validation and shorter time series

Interannual variability

±5.0%

Natural year-to-year weather variability at the project site

PV simulation uncertainty

±5.0%

Default modelling assumptions, simplified loss models, and hourly simulation resolution

Total PV yield uncertainty

≈±10%

Combined using the root-sum-square method

 

The lender

The bank reviews the project using the P90 energy yield.

  • With 13.5 GWh/year at P90, the project just meets the required DSCR of 1.25.
  • To stay safe, the bank limits debt to 70% of the project cost.

From the bank’s point of view, the project is financeable, but only just. There is no room for extra leverage.

The investor

The equity investor looks at the downside case.

  • About 4.9% P90 return on equity
  • The project works, but returns are modest
  • Capital efficiency is not great, but acceptable

The project reaches financial close, but it is not optimized.

Scenario B – “Let’s reduce uncertainty”

A company is developing a 10 MW solar PV project. The site looks good, the layout is clean, and the expected production is 1,500 kWh per kWp. On paper, everything works. 

The project engineer

Before finalizing the design, the team makes a different decision. They invest in better irradiance data, use longer time series, and apply more detailed modeling. The expected energy stays the same, but uncertainty goes down.

  • Expected P50 production still 15 GWh/year
  • Uncertainty reduced to ±8%
  • Expected P90 production increases to 13.8 GWh/year

 

Tab. 4. Uncertainty assumptions used in Sample Scenario B.

Uncertainty component

Value

Notes

Solar irradiance data uncertainty

±5.0%

Higher-quality irradiance datasets, longer historical records, and site-adapted validation

Interannual variability

±5.0%

Natural year-to-year weather variability at the same site (unchanged)

PV simulation uncertainty

±4.0%

More detailed system modelling, higher temporal resolution, and physics-based loss models

Total PV yield uncertainty 

≈±8%

Combined using the root-sum-square method

 

Nothing physical has changed yet—only the confidence in the numbers. With lower uncertainty, the project no longer needs to rely on conservative design margins to achieve the required P90 energy. While the installed capacity remains at 10 MW, the design can be optimized to reduce unnecessary conservatism, resulting in a lower total project cost.

Expected loss assumptions (such as soiling, temperature, and wiring losses) remain unchanged. However, improved data quality and modelling reduce the uncertainty around these losses. This allows a more efficient balance-of-plant design and lower cost without changing expected performance or relying on additional safety margins.

So the design is adjusted:

  • Same plant size of 10 MWp
  • Improved P90 energy compared to Scenario A: 13.8 GWh/year
  • Lower total project costs (a reduction of ≈1%

This is not cutting safety and durability. The project is simply no longer oversized to compensate for uncertainty.

The lender

The bank reviews the updated project.

  • P90 energy is higher than before: 13.8 GWh/year
  • DSCR improves to 1.29, so there is room for debt increase.

With lower risk, the bank is comfortable increasing leverage:

  •  Debt increases to 72%
  •  DSCR value keeps the same 1.25 as the previous scenario. 

From the lender’s perspective, risk has not increased, it has actually become better understood.

The investor

Now the investor looks at the downside case.

  • Higher P90 energy than Scenario A
  • P90 RoE increases to about 5.4%
  • P50 RoE increases to about 7.8%

An increase in minimum expected RoE was achieved, with less equity required for the investor.

Summary results

At first glance, both projects appear very similar. They deliver the same bankable energy and meet the same DSCR requirement. The site, technology, and market conditions have not changed. The only difference is how uncertainty is treated.

Uncertainty reduction is achieved by switching from the existing data and software setup to a more advanced solution. The additional cost of this change is assumed to be negligible compared with the potential capital savings, while increasing expected production at the P90 confidence level.

The comparison in the table below shows that actively reducing uncertainty changes how the same project performs. 

  • In Scenario A, higher uncertainty must be compensated through tighter lending constraints and lower capital efficiency. 
  • In Scenario B, reduced uncertainty narrows the gap between expected and conservative outcomes, strengthening the downside case used by lenders and investors and enabling more efficient financing decisions.

Further reductions in uncertainty could be achieved, if needed, through site-specific measurement campaigns and specialized consultancy services, offering additional opportunities to improve project efficiency and financial performance.

Tab. 5. Comparison of two uncertainty scenarios for a sample project.

 

SCENARIO A

SCENARIO B

Specific production

1,500 kWh/kWp

1,500 kWh/kWp

PV cost per kWp

$0.70

$0.70

Energy Price per kWh

$0.038

$0.038

P50 Annual Energy Production

15,000 MWh

15,000 MWh

Uncertainty (P90)

±10%

±8%

P90 Annual Production

13,500 MWh

13,800 MWh

Project Size

10.00 MWp

10.00 MWp

Total Project Cost

$7,000,000

$6,950,000

Loan Term

20 years

20 years

Interest rate

5.50%

5.50%

Debt (%)

70%

72%

Equity (%)

30%

28%

Debt ($)

$4,900,000

$5,004,000

Equity ($)

$2,100,000

$1,946,000

DSCR (P90)

1.25

1.25

RoE (P50)

7.6%

7.8%

RoE (P90)

4.9%

5.4%

 

Conclusions

How a PV project treats yield uncertainty has consequences far beyond the accuracy of a long-term energy forecast. The same expected production can lead to very different designs, financing structures, and investment outcomes depending on the level of uncertainty associated with that estimate.

Modern software solutions now integrate uncertainty directly into their calculations and allow it to be actively reduced through higher-quality inputs and more accurate models. This shift enables stronger technical optimization, more resilient financial structures, and greater confidence across stakeholders.

The impact of uncertainty reduction is no longer theoretical:

  • Projects with identical expected production can deliver very different outcomes depending on how uncertainty is treated.
  • Improved data and modelling can unlock optimization and capital efficiency, though benefits depend on project constraints and stakeholder acceptance.
  • Advanced data and modelling reduce uncertainty in PV yield estimates, but do not eliminate irreducible risks such as interannual climate variability or long-term weather trends.

Uncertainty reduction is a competitive lever, not a risk eliminator: it strengthens confidence and downside resilience, while broader market, grid, and regulatory risks remain decisive. Its impact therefore varies by region, as each market has its own critical factors, constraints, and financing conditions that shape how uncertainty influences project outcomes.

Reducing PV yield uncertainty does not change how much the sun shines. It does not eliminate project risk, but when applied effectively, it offers an opportunity to create a durable competitive advantage for the project.

Keep reading

Being certain about solar radiation uncertainty
Best practices

Being certain about solar radiation uncertainty

In the context of PV yield simulation, uncertainty helps users understand the potential deviations in the results produced by the software they are using. Understanding these deviations plays a key role in selecting the optimal design of a power plant and in evaluating financial risks and return on investment.

How to calculate P90 (or other Pxx) PV energy yield estimates
Best practices

How to calculate P90 (or other Pxx) PV energy yield estimates

One of the most critical outputs from PV simulations is the P50 annual energy yield estimate. Often referred to as the "best estimate," the P50 value represents the annual energy yield that has a 50% probability of being exceeded (with an equal 50% chance that the actual yield will fall below it).
However, relying solely on the P50 value may be too optimistic for project stakeholders. To address this, additional probability-based yield estimates are commonly used e.g. P90 value, which indicates the energy yield expected to be exceeded 90% of the time.

Introducing Solargis Evaluate 2.0: A new standard for PV project design and evaluation
Solargis news

Introducing Solargis Evaluate 2.0: A new standard for PV project design and evaluation

Today, we launched Solargis Evaluate 2.0, the next generation of our Evaluate solution. The new Evaluate is a more advanced, cloud-based platform for complete PV project design and evaluation, equipping project developers and industry experts with all the tools needed to tackle the growing complexity of modern solar projects.