America’ssolarindustry is wringing all possible efficiencies out of the project development process ahead of the 2016 federal Investment Tax Credit expiration. But are we missing an obvious hurdle?
Reliable project production estimates -- predicting how much electricity photovoltaic systems will produce annually -- are critical because they determine developer, seller and customer value. But what should be a simple process of estimating and agreeing upon production often turns into a frustrating and time-consuming “production estimate arm-wrestle” conflict during transactions.
Existing methods aren’t working, and with time dwindling to start projects before the ITC cliff, any delay in closing deals increases the risk a project won’t be able to capture federal incentives or remain competitive after ITC expiration. Here’s a look at the problem and a suggested solution.
Disagreement on production estimates can kill a project
Obtaining consensus on a production estimate is key to a deal, but the parties rarely agree, and the margin of disagreement may be 10 percent or more. This may not seem significant, but it can be the margin between a deal living or dying.
Production estimate disagreements typically occur between developers aiming to represent projects as highly productive and buyers or operators wanting to incorporate operational factors that will adversely impact production. One estimate may be excessively optimistic and the other excessively pessimistic, yet both parties are adamant their estimate is correct and are highly motivated to stand behind it, creating a time-consuming and frustrating impasse.
The solar industry needs a realistic and reasonable production estimation methodology. Decisions by project developers, system purchasers, system operators, investors, lenders, and electricity purchasers all depend on production estimates, which in turn depend on data inputs. These inputs should realistically reflect actual climatic conditions, as well as project design and operational details.
Problems with production estimate modeling systems
Today’s production estimates, typically based on “PVsyst” or similar software programs, provide many possible data inputs, which each party is strongly incentivized to manipulate to suit their bias and transactional goals. These influences may come from the inclusion, omission, overstatement or understatement of relevant inputs (intentionally or unintentionally), and the underlying input assumptions are difficult to detect but highly significant. Even the underlying climatic data can be manipulated.
In contrast, four or five years ago, our industry generally relied on the National Renewable Energy Laboratory’s (NREL) PVWatts calculator. PVWatts has the federal government’s credibility and impartiality and is simple to use. Available inputs were few and underlying assumptions transparent and plainly stated. Unfortunately, its simplicity and low price (it is free, after all) created the erroneous impression of low value and lack of accuracy, opening the door for highly customizable products like PVsyst, which have steadily become the norm.
These customizable products allow a wide variety and range of inputs, produce multi-page outputs, and do indeed create an impression of accuracy. But we should question the meaning of “accurate” when two modelers, using the same program but with differing assumptions, can produce results differing by 10 percent or more.
Data problems embodied by climatic modeling
This quandary is readily illustrated by climatic modeling. Consider an array in states where snowfall and snow accumulation are a regular annual occurrence. Unless manually cleared, the array will be covered by snow multiple days per year, and in many such locations, a full winter month’s worth of production will be lost in a typical year. Specific snow losses depend upon various factors including the array’s geographic location, the clearance height of the panels above the supporting surface, and the tilt of the panels from horizontal.
However, no prescribed guidelines exist to determine how much lost production should be attributed to snow, leaving individuals to decide this input. Developers will assume one figure (typically, zero) and prospective owners will assume other figures, which vary depending on their experience and conservatism.
Further, modelers can use their own climatic weather data rather than published and statistically validated data sets such as typical meteorological year (TMY) data from NREL. User-generated weather data may not be immediately apparent to the recipient of a production estimate, and such data, perhaps based on one year of measurements from a temporary weather station at the prospective array site, typically does not pass statistical tests for reliability and cannot be independently validated.
Production modelers typically claim user-generated weather data to be more accurate than published data because it is produced at the prospective array site. This cannot be verified, but notably, production estimates from user-generated weather data are only presented when the results are higher than from TMY data.
Decisions pertaining to whether and how to incorporate important production impact factors, such as having panels covered in snow for extended periods, are up to the modeler, and many people simply don’t account for them. Similarly, modelers can choose to ignore established meteorological data sets and selectively use self-generated meteorological data when the results favor the modeler’s goals. These two examples highlight how underlying factors in production models create disagreement, protracted contract negotiations, false expectations and failed deals.
Industry bodies should lead the standardization charge
Without better standardization of methods, the production estimation process will continue to generate time-consuming conflicts and slow deal closure, limiting system construction.
The solar industry will benefit by establishing standardized production estimating techniques, and its leaders should lead the charge toward publishing clear guidelines. Nationally recognized bodies like the Solar Energy Industries Association and North American Board of Certified Energy Professionals represent all sides, have the credibility to set a middle path, and are well equipped for the task.
This process could help negotiations now while smoothing the post-ITC development outlook. Countless aspects of a production estimate could be addressed through standardization, but any effort should at a minimum address:
- Meteorological data: Estimates should be based upon statistically significant and validated meteorological data sets using the data set closest to the site.
- Snow: Derate factor guidelines should be provided for snow states, ideally by county and with prescribed adjustments to differentiate between systems having low/high clearance heights and low/high tilt angles.
- Soiling: Guidelines should be provided, ideally county-by-county and month-by-month, but at least specifying minimum monthly soiling factors.
- Shading: The effect of horizon and near-field shading sources should be modeled, where present.
- Transformer losses: Losses from transformers located between inverters and production meters should be modeled (both operational losses and continuous parasitic losses). Guideline figures should be provided for high efficiency and standard efficiency transformers.
- Wiring losses: AC and DC wiring losses should be modeled, ideally based on actual design figures. Standard allowances should be provided for use when no design figures exist.
While greater production estimate accuracy may be achieved (a debatable point given difficulty confirming accuracy except over many years), the real benefit will come from standardizing production estimate methods, a reduction in opinion-led conflict, and ultimately, a greater number of operational solar projects.
We’ve no reason to continue arm wrestling
Whether or not the ITC is extended, America’s solar industry must become more efficient. Constructing more operational projects, faster and at lower transactional costs, are not idealistic goals -- they are both necessary and achievable today.
This is not a zero-sum game. Both sides benefit when unnecessary uncertainty, conflict, and related costs are removed from the equation. Any action to lubricate the development process is worthwhile -- unnecessary steps consume labor power and time.
Without standardization, the arm wrestling will continue and fewer projects will be built, but industry leadership could secure standardization almost immediately. While the numeric production results may not be demonstrably right, “right” also means closing more deals and building more systems in a manner that is better, cheaper, and faster than before. Standardized production estimation methods, prescribed by industry leaders and adopted by the industry, are an achievable goal. Let’s unite as an industry and achieve that goal.
Steve Goodbody is the senior vice president of operations and engineering of Soltage, a full-service renewable energy company that develops, finances and operates solar energy projects across the United States.