Existing solar people...how accurate did PVwatts end up being?

[petesamprs]

My installer saying PVwatts overestimates what a southern-facing, unshaded roof in NJ would produce. PVWatts saying 1400 KWh per KW, installer recommending a size that implies 1200 KWh per KW. Says in NJ he never sees anyone do better than 1300. What is your experience?

Looking to pull the trigger this week, but trying to pin down a final system size. Want to account for panel degradation for future usage needs, but don't want to overshoot.

[J.P.M.]

Basics: PVWatts is fit for purpose given what it's designed to do. What it does not do is handle shading at all.

To my experience, it's a good model for preliminary design of residential PV systems.

It's output is similar to other models I've used and similar to a model I designed about 12-14 yrs. ago.

What it is not is a predictor of system performance, particularly short term which is mostly dependent on the variability and chaotic nature of the weather.

If PVWatts inputs describe the system parameters reasonably well, AND the weather matches what PVWatts uses for weather in its modeling, the model will get within a couple % of what a system actually puts out.

PVWatts will and can be made to either under or overestimate a system's long term performance. GIGO. How close the model's output gets to any system's actual output is a function of the inputs to the model, how well the weather inputs match the actual weather and the period of time involved. Short time periods of less than, say 1 year can be off by ~ +/- 10%. Any modeled 30 day period's output is likely to be off by as much as +/- 30 % or so. Notice it's PLUS OR MINUS.

Your installer is either ignorant of how solar models work, particularly PVWatts, or, like most solar peddlers - at least to my experience - he's underestimating output to sell more product. Maybe he got his hand caught in the cookie jar when you challenged him on output.

No model is perfect. The PVWatts model, as stated above does not do any shading considerations. All models have their limitations and strong points. PVWatts is good for its stated purpose. BTW, it's a simplified version of something called "SAM" from NREL which is a pretty good design tool and does do shading. SAM is, however, not for the feint of heart.

Read all the PVWatts help/info screens a couple of times, then do your own input and make sure your model inputs match reality. I'd show your peddler this post and then get more quotes.

[Ampster]

I agree with the above. My installer also gave me a production guaranty and it was more optimistic than PV Watts. The first year my production exceeded both estimates. So far this year my production is slightly below those estimates. That only proves that weather is the biggest variable.
EDIT
43 Inches of rain this year
20 inches last year

[solarix]

I find PVWATTS to be quite accurate - at least for my area. I find many solar dealers overestimate what their systems will produce to make their ROI look better and convince the customer what good value their product will be. I instruct customers how to use PVWATTS so they can see how the dealers are trying to scam them.

[petesamprs]

I have the opposite problem. My installers are saying that PVWatts tends to *overestimate* what will actually be produced, so they are recommending larger systems than would be implied by PVwatts. I know PVWatts doesn't take shading into account but my house has no shading at all on the south side, so that can't be the reason.

[ButchDeal]

I have the opposite problem. My installers are saying that PVWatts tends to *overestimate* what will actually be produced, so they are recommending larger systems than would be implied by PVwatts. I know PVWatts doesn't take shading into account but my house has no shading at all on the south side, so that can't be the reason.

many installers just do a simple rule of thumb to come up with the production values. Some even use a wrong rule of thumb... I found one that had a bad compass, his measurements were always off by ~20 degrees. He was a good installer but get a smart phone or new compass.
Others just fudge the numbers up to either make sure they have enough production or increase sales.

Even the ones that actually use the simulation models are often silly about it. They don't always put in the shadows correctly (trees, other buildings etc).
They are all just tools, even the rules of thumb, and they are only as good as the user using them.

[Elemental101]

I did a PVWatts and it got it right for me but on the lower end of it's estimate but I did change the estimated system size.

[J.P.M.]

I have the opposite problem. My installers are saying that PVWatts tends to *overestimate* what will actually be produced, so they are recommending larger systems than would be implied by PVwatts. I know PVWatts doesn't take shading into account but my house has no shading at all on the south side, so that can't be the reason.

Of course all, or at least in my experience most, PV peddlers will claim most any model will overestimate output. That's how they justify B.S. low estimates for output. Then, those low estimates make is easier to make solar ignorant marks believe the peddler's B.S and buy oversized and thus less cost effective systems that, among other drawbacks, most likely take longer to achieve financial breakeven.

Look - a dose/opinion of real world logic: Solar outfits make money by putting PV on your property - not saving you money. No salesperson I know of (and I was a commission salesperson for ~ 10 years peddling industrial power and process equipment and took a pay cut to change to engineering) ever got fired for selling too much. What would you expect them to do ? Besides, selling real and optimized/maximized cost effectiveness reduces the seller's profit. Seems to me PV sellers have no real incentive to not oversell.

Respectful suggestion: Reread all the PVWatts info/help screens. Then read "Solar Power Your Home for Dummies". Make sure your inputs for PVWatts are close to reality for you application and do another run for a 1 STC kW system. Then, ratio that size up for how many kWh/yr. you want to offset with PV divided by that 1 STC kW system's output. With no shading, you'll be close within probably +/- 10 %/yr. or closer due to annual weather variation, and over, say, 10-20 yrs. you'll probably be within a few % of a pretty god est. of system size.

Show this post to peddler's who try to B.S. you before you run them off.

Take what you want of the above. Scrap the rest.

[DrLumen]

MIne currently is running about 2mWh less per year than what pvwatts and the contractor predicted. ~13-14 mWh predicted but 11.6 actual.

However, much like Ampster, N. Texas is in the 100 year or 500 year flood years so lots of clouds and rain.

[NCSD]

I've found PVWatts to be useful for rough estimates. The only major issue I've observed was in the hourly estimates. The PVWatts hourly estimates do not appear to use DLS, so I needed to shift results by ~1 hour for some months. The specific hours are relevant for me rather than just total generation per year since I was using PVWatts to decide which SDG&E Time Of Use plan to choose, which depends on how much energy is generated in specific hours. I needed to make some other minor adjustments as well for best results. I expect some of these relate to nuances of my specific setup, and others differences between weather/sunrise/sunset of my location and the PVWatts source data.

[J.P.M.]

MIne currently is running about 2mWh less per year than what pvwatts and the contractor predicted. ~13-14 mWh predicted but 11.6 actual.

However, much like Ampster, N. Texas is in the 100 year or 500 year flood years so lots of clouds and rain.

How long has that period of analysis been ? PVWatts suggests the model might be +/- 30 % over 30 days. +/- 10 % or so over a year.

Writing anecdotally, my system's PVWatts modeled production is 9162 kWh after consideration for 3 % fouling and 3.5 % shading. Actual running 365 day production since 10/17/2014, with startup 10/17/2013 is 9091 kWh/365 day period. (min. = 8,716 kWh, max. = 9,571 kWh, std. dev. = 194 kWh). The range of - 4.1% to + 5.3 % is tighter than the PVWatts claim of +/- 10 %.

For running 31 day periods since 31 days after 10/17/2013 the ratio of actual 31 day production vs PVWatts modeled production for the same periods:

Actual 31 day production/PVWatts modeled 31 day production for the same 31 day period: average= 0.9893, min. = 0.6019, max. = 1.233mean, std. dev. = 0.0998.). That std. dev. for my data suggests the monthly variation at a 99 % confidence level of ~ +/- 25 %, also inside the PVWatts claimed +/- 30 %.

[josefontao]

For me, PVWatts overestimated. Here are my numbers. On the top is what I've been generating, the bottom numbers are the pvwatts estimates.
solar generation.JPG

[Ampster]

Some of it could have something to do with the weather. Although that might even out over time. Also check the assumptions you put into PVWatts.

[J.P.M.]

For me, PVWatts overestimated. Here are my numbers. On the top is what I've been generating, the bottom numbers are the pvwatts estimates.
solar generation.JPG

PVWattts is only a model. It's no more accurate than a long range seasonal weather forecast. Think of it more as you would a climate indicator than a weather forecaster. Its output will be ~ +/- 10 % or so over any year. Something like +/- 30 % or so over any 30 day period. Better than that is luck/randomness. Check the PVWatts help/info screens for confirmation/further information/confirmation.

[Salts]

I designed my own solar system and used PVWatts to help me figure out how much solar I would need to meet my requirements. We took our utility bills and went back 5 years crunching data to figure out how much solar we'd need to take care of the entire home and then designed the system for those requirements.

The annual production numbers PVWatts gave me were within 2% of what I actually generated the first year. The second year was about 3.5% but the weather has been unusual.

We have a 100 foot maple tree located on south side of the array about 120 feet away and slightly offset to the array. During the summer, its not an issue, but when the sun goes low to the horizon during the winter, the top of the tree casts some shadows onto one side of the array for a few hours each day and it drops the power that string is making. I did not think the tree would be that much of a problem. It ends up costing me about 3% of my annual power generation so it looks like PVWatts is spot on.