Operational decision-making at scale has never been easy, and the sometimes-fine line between taking preemptive action and the cost of potential loss can be paralyzing when decision-makers lack good data and/or analysis. Without robust data and an analysis tool, deciding when to spend money to prevent losses becomes a complex decision. And the timing on this kind of decision is crucial.

Here we step through a real-world example of applying probabilistic S2S weather forecasts from Salient Predictions to a weather-dependent decision: if / when to clean solar panels in the desert.

The Challenge

• Solar panels in the desert get soiled by dust in the air and pollution
• When soiled, panels are less efficient – electricity production goes down, and asset owners lose money
• Soiled panels can and need to be cleaned, but cleanings are expensive
• What happens with the weather complicates matters: rain in the desert cleans panels fairly well, but if it doesn’t rain, asset operators need to schedule and pay for cleanings
• And, frustratingly, what if an asset owner pays to clean panels and then it rains?! Wasted money

The Solution

Given that dirty solar panels are less efficient and cleanings are expensive, when and if to spend money cleaning a solar farm to enhance the efficiency (and revenue) is critical to the financial health of a solar generation asset.

Solar Unsoiled optimizes the cleaning schedules and frequencies of solar farms through predictive modeling tuned to each site using analysis of site data and microscope images of the particles on the solar panels. A key input to this predictive soiling model is Salient’s seasonal to subseasonal (S2S) precipitation forecasts. Salient’s reliable, accurate probabilistic weather forecasts help decision makers make data-driven decisions on 2 week to 1 year time horizons.

With the combined technologies, the decision on when and how often to clean soiled panels is answered, greatly reducing the risk of wasting money cleaning right before a rainstorm.

A Data-Informed Approach

Solar Unsoiled’s modeling and analytics predicts and advises when it is economically optimal to clean solar farms. Salient’s probabilistic precipitation forecast is then used to quantify the likelihood of a rain storm in the weeks ahead that would render the predicted cleaning unnecessary.

The Case Study

How does the addition of Salient’s 2-week precipitation forecasts affect cleanings and profits? To answer this central research question, we selected a single, 25 megawatt solar farm in Arizona for this analysis, and modeled three approaches to cleaning over 8 years:

• Approach 1: no cleanings
• Approach 2: clean once per year in June
• Approach 3: use Solar Unsoiled’s optimized cleaning schedules powered by Salient’s S2S precipitation forecasts

The basic steps of the analysis included:

1. Establishing site parameters:
   1. Location: Tucson, AZ (32.3, -110.0)
   2. Size: 25 MW
   3. Cleaning Cost: $18,500/cleaning
   4. Analysis period: January 1, 2016 - December 31, 2023
2. Establishing Solar Unsoiled’s optimized cleaning schedule for the site
3. Determining a probability threshold that defined a >4mm rain event (enough to clean a panel) from Salient’s forecasts
4. Creating a Salient-and-Solar Unsoiled optimization
5. Comparing profits of the three soiling profiles

The conclusion: Approach 3, the data-driven approach that combines the two technologies to make when-to-clean decisions, generates ~$28,000 more in profit per year over Approach 1 (no cleanings) approach, and ~$16,000 more per year over Approach 2 (cleanings once per year on a predetermined date).

In context: Utility-scale solar is a low margin, high volume business. A 25 MW site like the one profiled sees an annual revenue of ~$2.5 million, with ~10% margins. Addressing soiling using a data-informed approach like this one translates to ~11% higher profits for solar asset owners ($28,000 in additional profit on $250,000 margin).

Why this matters:

This is an analysis for one farm. Extrapolating across the American Southwest which is home to ~80 GW of America’s 180 GW of installed utility-scale solar, solar farms in the region stand to make an additional $66.7 million in profit per year.

For a heavily subsidized industry like big solar, small changes in margin can make or break a financing deal. As the managing director of one solar development company explains, “having access to soiling estimates and accurate yield predictions are crucial for the project because the project’s financial pro-forma estimates are critical. If the site revenue estimates aren’t appropriately adjusted for soiling, the project will look like it can afford more cost than it actually can.” Given the tight margins, 11% profit improvements and an additional $66.7 million in the region is meaningful.

The benefits of higher margins like these accrue to a range of stakeholders. Asset owners and investors see higher returns and increased asset value. Solar farm operators unlock higher levels of operational efficiency, which can bring performance bonuses. Utility companies have access to stable, cost-effective renewable energy supply that helps them meet regulatory compliance requirements. Efficiency increases eventually drive down energy prices for consumers, and profitable solar farms contribute to job creation and increased tax revenues for local communities.

And, of course, enhanced solar farm performance contributes to reducing greenhouse gas emissions in the face of a changing climate, helps us meet our regional and national renewable energy targets, and encourages further investment in renewable energy infrastructure.