Solar PV Performance Analysis: 3.3. Clipping

Clipping refers to the scenario where inverter output is maxed out and flatlines even though irradiance continues to increase. This causes "losses" as the DC field can generate more electricity but the inverters cannot as they have reached maximum capacity.

Think of a chart like the one below as a prime example of clipping, we see the irradiance continues to climb whilst the inverter output flatlines (is “clipped”):

Clipping is influenced by the DC/AC ratio, which is the capacity of the solar modules (DC kWp) with respect to the capacity of the inverters (AC kW). E.g. if I have 12,000 kWp of solar modules and 10 x 1 MW AC inverters, the site has a DC/AC ratio of 1.2.

DC/AC ratios can be any number but are typically between 1 to 1.5.

In higher latitudes (north or south hemisphere), DC/AC ratios tend to be higher as plant designers look to boost production in winter when the sun angle is lower. But of course the higher the DC/AC ratio, the higher the plant cost (more modules and cables etc), so DC/AC ratio design has to consider both increased output vs increased cost.

In PV Syst the clipping losses are defined as “Inverter Loss over nominal inv. Power”.

So when looking into the performance of a solar farm, we’re comparing the actual losses from clipping vs the modelled losses in PV Syst. For example, if the PV Syst simulation for a solar PV power plant has 0% losses for clipping, then any actual clipping losses are underperformance. If the PV Syst simulated has 1% losses for clipping and we calculate actual losses are 1%, then there is no underperformance.

Grid curtailment can also look like clipping, so when analysing clipping losses it’s important to remove any periods when there is a grid curtailment order or zero export limitation. And we should also remove any other outages (e.g. inverter shutdowns) so that we isolate pure clipping losses in our calculation.

Calculating clipping losses when at least one inverter is not clipped

To calculate clipping losses when at least one inverter is not clipped, you can simply compare the output of clipped inverters (pro-rating output if inverters have different DC and/or AC capacity) to the output of non-clipped (“reference”) inverters on the same site.

E.g. if a non-clipped inverter produces 1000 kWh per day and the clipped inverter produces 900 kWh per day, assuming there are no other outages or loss differences between the inverters and they have the same DC and AC capacity, then the clipping losses are 100 kWh or 10% for that clipped inverter. Then to get a better idea of overall clipping losses, this exercise needs to be done on multiple days across the whole site.

Calculating clipping losses when all inverters have clipping --> Regression

If all inverters are clipped then the above method doesn’t work because there are no reference inverters. In this case we need to use a simple regression model (y = mx + b) based on the relationship between irradiance and output on that site.

Comparing the irradiance and the output of the site during non clipped periods allows us to determine a regression formula (slope + intercept) in order to work out for any irradiance level, what is the predicted output. We can then use this predicted output to evaluate the clipping losses (again assuming all other loss sources have been eliminated from the calculation).

In the above example, clipping starts around 08:00 and continues until around 15:10. We can therefore use the correlation between the irradiance and output before 08:00 and after 15:10 to obtain our regression formula, which can then be applied to the periods when there is clipping to calculate the expected output.

Again as with the other sources of losses we have looked at, it’s important to track clipping losses to compare against the financial model and as a feedback loop for future projects. E.g. if clipping losses are too high then for future projects of similar design, a lower DC/AC ratio could be favoured.