Friday, February 14, 2014

Grid Tied Inverter Overloading Analysis (with PVSyst)

For today’s post, I brought out the solar modeling software and gotten myself into trouble (grid tied inverter loading analysis).  System analysis and solar modeling software is a delicate subject, there are dozens of variables and different models can produce different results.  I've tried to keep things as simple as possible for the purpose of this analysis.

When you hear about inverter loading/overloading, it refers to the ratio of Solar DC to Inverter AC power (Watts) that you design into your system.  This can be expressed as a percentage value.  For example, a system that has an inverter that’s “20% overloaded” (or 120% loaded) would mean the DC array size is 20% larger than the AC rating of the inverter.  A system that is 0% overloaded (or 100% loaded) would have a solar array and inverter that equal each other in size.  For the purpose of this analysis, I created three different inverter loading scenarios using Power One string inverters and Canadian Solar panels.

Inverter
Solar Array
Inverter Load
Power One 6000   (6 kW)
(24) Canadian Solar 250W Poly
2 Strings of 12
100%
Power One 5000   (5 kW)
120%
Power One 4.2     (4.2 kW)
142%

It’s important to keep the string size consistent when making comparisons like this to cut down on loose variables.  In this case, I was lucky to find a situation where an identical solar array can be compared across three different inverter loading scenarios in a fairly linear fashion.  The only difference between the inverters is a slight decrease in the CEC efficiency (0.5%) from the 5 & 6kW inverters to the 4.2kW version, while maximum efficiency and voltage windows are almost uniform across all three.  I corrected for the CEC efficiency loss by increasing the 4.2kW inverter production numbers by 0.5%, but in the end this had little to no effect on the data.

The different loading ratios have been simulated across four different cities for average annual production:

Albuquerque, NM – A good representation of a dry climate with high-intensity sun.

Miami, FL – A sunny site with interesting weather effects and a lot of humidity.

Chicago, IL – A colder site that gets less sunshine with long cloudy winters

Kansas City, MO – An average site that has characteristics of the other three sites.

To estimate production, I’ve used PVSyst (V5.65) for the simulation software.  The system orientation was set at a 30 degree tilt and true south azimuth.  The following results were obtained for average annual production across all sites and loading conditions:

PV SYST PRODUCTION (kWH)
Inverter Size:
6kW
5kW
4.2kW
Site:
kWH/yr
kWH/yr
% Loss
kWH/yr
% Loss
Albuquerque, NM
11191
11138
0.47%
10655
4.79%
Miami, FL
8866
8866
0.00%
8796
0.79%
Chicago, IL
7998
7998
0.00%
7878
1.50%
Kansas City, MO
9071
9066
0.06%
8877
2.14%

Production numbers are to be expected for the four cities.  Some may be surprised that Miami is so low while Kansas City is higher.  The reason is Miami has a lot of humidity and a fairly uniform temperature profile year-round.  Kansas City has hot summers, but their winters are fairly cold and that boosts solar production, and their dryer climate helps more of the sun’s rays hit the solar array.  Chicago has the worst solar production as expected, but they are affected more by inverter loading than Miami.  This is due to colder overall temperatures allowing the solar array to produce more continuous power during optimal conditions, which can cause clipping on an overloaded inverter. 

What jumps out immediately is the change in production in going from a 6kW inverter to a 5kW inverter, or lack thereof.  There is virtually zero difference in production for three of the sites by going to an inverter loaded 120%, and for Albuquerque it’s only half a percentage point off the top.  This is negligible, and after the first year or two of service there will be no difference due to degradation of the solar panels (which typically lose ~0.5% of their production per year installed).  It can be seen that overloading the inverter by 20% is something that should be designed into any solar job, while the benefit from matching the inverter rating with the solar array size just isn't there.

The 140% loaded inverter has more significant production losses than the 120% inverter.  For Albuquerque especially, the difference is too large to be ignored, and overloading the inverter that much should not be attempted at similar sites for optimal orientations.  For the other three cities, the power loss is pretty bad but there is still hope.  The simulation parameters are for a system tilted at 30 degrees, oriented optimally with respect to the sun, all with zero shading.  If a system is being considered with sub optimal orientation or shading, anything that might adversely affect production, an inverter loaded to 140% isn't out of reach and in many cases will work just as well.  The production difference is only 0.8% for Miami, so temperate sites without cold winters could have 40% overloaded inverters without any trouble.


What are the biggest takeaways from this?  The better the site is for solar (days of sunshine, low humidity, cold winters), the less the inverter should be overloaded.  The minimum for inverter loading falls around 120% for the United States.  Anything lower and you are wasting inverter capacity, anything higher and you need to consider the site conditions before proceeding.  Temperate climates where winter temperatures do not drop below zero have the most options with inverter loading, and 40% is not out of the question.  Climates where clear and cold conditions occur have to be careful in over sizing the solar array for the inverter past 20%, colder temperatures let the solar array produce more power and can cause production losses if the inverter isn't sized correctly.  It is this designer’s opinion that 120% inverter loading will work just about anywhere, and should be the standard when pairing a solar array with an inverter.

No comments:

Post a Comment