Too much of a good thing: Inverter hyper-clipping

By Paul Grana, co-founder, Folsom Labs

In earlier articles we’ve already pointed out that inverter clipping isn’t as significant as most people think, and that in a grid-power-constrained system it may be economically optimal to have a rather large DC-to-AC ratio. But what if there was a situation that resulted in significantly higher (and unexpected) inverter clipping losses of 30% or more? We’ll show you how to understand and avoid these cases of inverter “hyper-clipping.”

What really happens when inverters clip power?

It’s easy to say that the inverter “clips the excess power,” but from a physics point of view, that doesn’t describe what is going on. You can’t just “throw away” power you don’t want—and inverters don’t have air conditioners they can turn on when they need somewhere to send the excess energy. The main tool an inverter has is setting its DC voltage—and this is actually how an inverter is able to drop the system power.

It helps to visualize this issue graphically. A solar array has a power-voltage curve that illustrates the relationship between the operating voltage and the array’s output power. The modules can perform anywhere on the curve, and it’s the inverter’s job to pick the spot on the curve—ideally at the spot that maximizes the power (called the max power point, or MPP).

Figure 1: Typical array power-voltage curve

At the same time, an inverter has a maximum operating power and a voltage range it operates within. We can visualize the inverter’s operating range as a rectangle.

Figure 2: Array power-voltage curve in over-power clipping

When an inverter is in an over-power clipping mode, the array is producing more power than the inverter can handle. The inverter will increase the DC operating voltage, pulling the modules off of their max power point, until the modules’ DC power is within the inverter’s operating range. You can see this as the green point in Figure 2. The inverter protects itself while maintaining maximum power production. The modules end up dissipating the excess power as heat, but as we’ll show at the end of the article, this isn’t a big deal.

However, there is a scenario where this behavior can cause problems. Specifically, look at what happens if the arrays’ power-voltage curve doesn’t intersect the inverter’s operating range. The process we described above (the inverter increasing the operating voltage until the modules’ DC power is within the inverter’s operating range) doesn’t work. Instead, the array will miss the inverter voltage window and trip off—for that period of time, the energy production will be zero. Note that this is by definition happening at a time when the array is at peak production—so just a few times a year can have a serious impact on the array’s energy production!

Figure 3: Array power-voltage curve in over-power and over-voltage condition

So how might this come up?

The description above is a theoretical framework, but how might this issue come up in an actual system?

There are a few ingredients needed to make this happen: a location with lots of sun (high power) combined with relatively cold temperatures (high voltages), high designed string voltage relative to the inverter’s max operating voltage and a large DC-to-AC ratio.

We can look at the power and temperature properties for a few locations around the US, and it looks like there are a few cities that are at elevated risk for this (again, high irradiance relative to temperature). We’ll use Los Angeles for the analysis here.

Figure 4: Sunlight and temperature by location

We then design a system in Los Angeles with an inverter with a max MPP voltage of 750 V (combined with a 1000 V Voc string), and 1.5 DC/AC ratio.

Figure 5: Loss chart for hyper-clipping simulation

We can see that the clipping losses can be as high as 32%, caused by 15% of operating hours where the array goes into hyper-clipping and trips to zero.

In terms of seasonality, these clipping losses persist for most of the year, with clipping losses above 30% for seven months of the year.

Figure 6: Clipping losses by month

Sensitivity to design choices

It’s worth illustrating how these two factors interact. Note that if we start with a base case of an array with a 1.2 DC-to-AC ratio and an inverter with a wider max voltage of 820 V, then there is no clipping loss. Each factor independently will lead to clipping of 5.7% (for increasing the DC/AC ratio to 1.5), and 0.6% (for dropping the inverter’s voltage to 750 V. But together, the clipping losses jump to 32.4%—approximately five times the sum of the individual effects.

The good news is that HelioScope will properly simulate these losses—so if you are ever at risk of hitting these conditions, you’ll find out before you build the array. And it’s a reminder that inverters aren’t just black boxes that turn DC power into AC power. The nuances of their behavior (including the operating voltage range) can make a big impact on energy yield.

How much heat does this create at the module?

This description of clipping often raises questions about the module health. Basically, if the inverter isn’t ‘clipping’ excess power but the modules are, then does this damage the module?

To re-state the process described above: During inverter clipping, the modules are working off of their maximum power point. So at a moment when a module wants to produce, say, 320 W, it is only able to deliver 240 W to the inverter. The difference (in the example, 80 W) results in heat at the module. So, how much heat are we talking about?

Here, it helps to think in terms of thermodynamics. Modules are only about 20% efficient at converting sunlight into energy—and the rest, 80%, is largely dissipated as heat. So, say an inverter is clipping 25% of the array’s production (as in the example above). Then in the broader context of the sunlight, the 20% efficient module is only converting 15% of the sunlight’s energy to power, with the resulting energy, 85% of the sunlight, being converted to heat. Compared to the base case (80% of sun’s energy converted to heat) this is an increase of ~6%. Sure, any extra heat isn’t ideal—but any well-made module should have no problem handling that extra heat.