Is it worth it to wash a dirty array?
When I get a few questions about a certain topic, it is probably worthwhile to explore it since there are probably more people with the same question. I had nothing better to do while I babysat this smoker full of ribs so I decided to write about it. Solar array washing services have popped up in all major markets. I saw this neat auto washer at RE+ 2022 in Anaheim.
I was curious about this service since I am a bit old school and cling to the idea that a good rain will wash the array. However, the last 10 years here in California have got me doubting this premise. Wildfires and the lack of rain have turned most arrays in this state to a dullish-brownish color. Here is what my array looked like before we got all this rain.
Gross, right? A few months ago, a lady on NextDoor made a post stating she paid $250 to have her 8-module array professionally cleaned. I could not wrap my head around this for such a small array and the person who washed the array was quick to jump in. He compared this washing to an oil change, which I thought was a stretch. I asked if he presented any production estimates for his service and that was the last I heard of it.
Not to let a training opportunity go to waste I decided to compare my production pre and post-California rains. This is not as easy of an exercise as it seems.
“Huh? Just compare a few days before the rain to a few days after. Duh…”
Well, nothing is ever that easy in this business that relies on the mercy of Mother Nature and that gas ball 93 million miles away. And let’s not forget the tolerances of the modules and all the other things that go with PV production estimates. I do not have access to powerful estimation tools like PVSyst or Helioscope that may be able to estimate these gains (or losses) so I will rely on some barbaric math and SWAG. The days leading up to rains with good sun were in December and the rains subsided in the middle of January. There are many variables to consider, but fortunately, the weather is fairly constant here in CA so it made part of this exercise easy.
I found 6 days in December that were bright and sunny that lead up to the rains. I picked the next 6 sunny days in January to compare. Luckily, we keep accurate records of weather conditions via airport logs. The dates I chose to compare came from my SMA Sunny Portal charts. I have a 5000W inverter with 5120W of Jinko 320W modules in two strings of 8.
I will geek out a little here since I am a big fan of citing sources- something some solar companies rarely do when they make bold and rash statements. Here is the monthly data for December 2022:
I will focus on the sunny days December 12 – 17. I also found the temperatures for Sacramento International Airport (SMF) for those dates. Here is the production data from the chart compared against daily high and low temperatures. The total production for those days is 93.44 kWh. You can see that the temperatures are fairly constant. The average high and low is 53/30 respectively. This is important because temperature affects PV array voltage.
High / Low
The daily variable that is difficult to find is irradiance- the amount of sunlight per square meter that strikes a given point. There are irradiance charts that show what it should be but not what it was. That’s ok, PV Watts gives a decent estimate. For my area, the average irradiance for December is . January is a tad higher. Since irradiance has more of an effect of current, this is an important variable to consider because (voltage x current) x time = energy (kWh).
January 20 - 26, 2023, my next six sunny consecutive days, reveals a bump in production.
We see a 17.9% increase, but let’s bump it to 18%. Some may be tempted to stop here. “Well, there it is. The rain did help production!”
Not so fast- remember, there is more irradiance in January than in December. The average temperatures for January were also just a smidge higher. Also, those same sunny days in January 2022 show production was 106.41 kWh which is 3.4% lower than the previous year. PV systems usually lose production year after year. See how difficult this exercise is?
But not to be a total curmudgeon, let’s take the bump in production at face value and throw caution to the wind!
The difference between the two sets of values is 17 kWh over six days, which comes out to 2.83 kWh per day- for a range of temperatures and a range of irradiance. For reference, a refrigerator consumes 2 kWh per day. To make numbers easy, we’ll bump that up to 3 kWh. Sacramento Municipal Utility District (SMUD) is on a time-of-use schedule like most of California. We will use the off-peak rates since there isn’t an appreciable amount of solar past 3 p.m. this time of year. Lot’s of assumptions in this exercise as well.
Ok, so using the off-peak rate (at $0.11 kWh x 3 kWh), we arrive at a modest savings of $0.33 per day. Our 8-module array customer would need $250 / $0.33 = 757.57 days, just like those in January, to recoup her washing fee. It will probably rain again before that time is up (even here in California). It doesn’t seem worth it to me. I have a twice as big system, so it may cost twice as much if we use $/module. But everyone has their own definition of value and what makes them sleep better at night.
I made a blog promo for this post on my TikTok page and one person had an interesting comment about his true-up after he cleaned his array 3 times in 3 years. And to be totally fair and transparent, I did not consider the increased irradience and higher production values for the summer months, which may reduce this payback time even more.
For larger systems, say a 250,000 kW commercial rooftop or a 100MW utility-scale project under a Power Purchase Agreement (PPA). An 18% increase in production for any amount of time for these larger arrays is worth a lot of money to the investors of these systems.
So what’s the real answer to, “Is it worth washing my modules?” is the same as most questions in this industry.
Time to wrap those ribs.
*The author was babysitting 2 racks of ribs on his smoker and enjoying adult beverages while writing this blog post. Please excuse any math errors.