Domain Knowledge Bite: Clean irradiance data – without it, we’re working in the dark

Clean irradiance data

Whether you are the owner of a solar plant or the operator, understanding its performance is key to success. If you don’t know how well a plant is performing, then you can’t budget for the future, and you can’t make improvements.

Measuring performance sounds straightforward. The first step is to get a reliable measurement of the irradiance, because it’s the irradiance that drives everything else. Once you know that, then you can calculate how much energy a perfectly performing plant should be producing – any shortfall is a loss, and quantifying losses is the heart of measuring performance.

As Clay Helms, Performance Engineer III for Silicon Ranch says… “At Silicon Ranch, we monitor, analyze, and maintain our irradiance data as seriously as we do our equipment. As the foundation to nearly all understanding of plant performance, high-quality irradiance data is critical to the long-term ownership of healthy solar power plants.”

Every solar park of any size will have sensors installed to measure irradiance throughout the day. But what happens when a sensor fails, or goes out of calibration? Then all you can say is “the sun was shining, and some power came out…” Not much use when you’re planning next year’s annual operating plan. This problem is more common than you might expect: in a recent study for a customer, we found about 5 % of irradiance data to be unusable. What is needed is a way to clean this data, and replace it with a reliable estimate.

Fortunately, there is a way. Most parks have multiple sensors, and data from the working ones can be used to replace the missing data from a faulty one. The trick is to recognise when data is bad – that’s not always obvious when a sensor is just drifting out of calibration – and then to combine the multiple sources of good data to give the best possible estimate of a correct reading. As part of our Renewables Optimisation mission here at GreenPowerMonitor, we’ve been putting a lot of effort into finding a way to do just that, with very satisfying results.


In the chart above, the dotted red line shows data from a faulty pyranometer. It’s on a site where panels, and the pyranometer, are mounted on trackers. In this case the tracker has become stuck in the stowed position, so the sensor isn’t reporting the required plane-of-array data. Our algorithm has detected it, and replaced the signal with corrected data from other sensors on the site.

This cleaned irradiance data opens the door to reliable performance reporting. Regular reports, be they monthly or annual, go to the many different stakeholders of an asset, and it is essential that they can be trusted. Knowing that they are based on a reliable source of data, gives the reader the confidence to use them as the basis for critical decision making.

Once irradiance and expected production have been established, a good analytics system, such as GPM’s SolarGEMINI, will quantify the lost production: the difference between the expected production and the amount of energy which is actually produced by a plant. And SolarGEMINI will go further than that, finding the losses due to each of many possible root causes. That’s invaluable, because it tells you which losses are unavoidable, such as grid outage, and which an operator has some control over: inverter outage, for example. Once they are understood, losses are classically displayed as a waterfall chart:


irrdiance 2

To an asset operator, a breakdown of losses provides invaluable guidance. At a strategic level, it shows how successfully the plant is being maintained, and where there is room for improvement. At a day-to-day level, SolarGEMINI will drill down into the data and discover the likely cause of a loss: there was no scheduled maintenance… the plant wasn’t clipped or curtailed…there is no more shading than usual…the inverters are functioning correctly…but string combiner box #327 has failed. Bingo!

Cleaned irradiance data has other uses beyond performance measurement too. It can boost the accuracy of production forecasting, helping to maximize the return from energy trading. And it is used in preconstruction assessments of new projects: an accurate understanding of the climate is essential to predicting the energy yield of a future solar plant.

Having clean, reliable irradiance data gives a great boost in the challenge of understanding solar plant performance. It is the fundamental building block on which all analysis rests, and gives owners and operators the starting point from which to make decisions with confidence.

Categories Noticias @en