[RE-wrenches] insolation v. actual output

[Matt Lafferty]

Marco,
 
Are you looking for an instantaneous AC power output number based on
equipment configuration and environmental conditions? Accurate to within a
few percent one way or the other? To have a general idea if the system is
performing properly or not?
 
You need a lot more than just irradiance to figure it out. Energy generated
over time is generally a much better metric for customer use than
instantaneous power output.
 
The basic process requires determining an "Expected" output and comparing it
to a measured or observed value. In our case, we are looking for watts. AC
Output Watts specifically. Can't stress enough how important it is to manage
customer expectations. Would love to hear how the customer ever heard of a
PTC rating in the first place, and how they came to think it would be
representative of the power output at the inverter... I am of the strong
opinion that we should present conservative expectations. Under-promise and
over-deliver. 
 
Part of managing expectations successfully includes establishing an
acceptable range of accuracy tolerance. The California Solar Initiative
requires a Field Verification of performance that has pretty wide
tolerances. You might want to use this methodology with your customer. CSI
forms are available from the program administrators (utility company).
PG&E's version is here:
http://www.pge.com/includes/docs/word_xls/shared/solar/csi/form_fieldverific
ationcertification.xls
 
To use the form, multiply the STC DC rating of your system by the factor
found in the table for your measured plane of array irradiance and ambient
temperature in degrees Farenheit. As long as your inverter display says the
system is putting out at least that value, it's "acceptable". In my opinion,
this is fine for laymen but far too loosey goosey for any real technical
evaluation or diagnostics. User Note: Do not attempt to use this method when
any part of the array is shaded.
 
SolarPro 2-6 (Oct/Nov 2009) had an article on Commissioning that had a
decent rundown on the process from a commissioning/acceptance perspective.
(Same article that Jamie J is talking about). You are looking for the
process described on Page 52. If you are planning on providing your customer
with this formula so they can figure it out any time they want and stop
bothering you... I would tweak it a bit so it is more useful on a regular
basis. So they actually stop bothering you sooner.
 
The basic formula that you are looking for is this:
 
Minimum Expected Power Output = Total STC Watts * Irradiance Factor *
Temperature Factor * System Derating Factor
 
This formula includes variable and constant values. The System Derating
Factor is made up of several factors, some of which are variable and some of
which are constant. And therein lies the rub for the purpose of handing it
to your customer and hoping never to hear from them again unless there is a
real problem. As opposed to a perceived problem, that is.
 
>From my standpoint, I don't want potentially irrational customers with
scientific calculators trying to estimate a shade factor for instance.
Unless there isn't any shade. Same goes for soiling. 
 
So, back to coming up with a formula that you can hand them so they can run
back and forth between the roof, the inverter, and their computer. The key
here is, well.... A couple things toward expectation management.
 
First and foremost, a single measurement does not a comprehensive evaluation
make. Ain't happenin'. You (they) need to measure, observe, and record
multiple measurements, over time, to gain a real understanding of the
performance characteristics of a given system. Trends matter. A lot more
than a single point-in-time measurement. If your customer is an engineer or
accountant, tell them it's Statistical Process Control and they'll be happy.
 
Secondly, point-in-time instantaneous measurements are subject to changing
environmental conditions. Also known as "the math don't always add up
syndrome". You need to take all your measurements and observations at
precisely the same instant. This is harder to do than it might seem.
Particularly when the array is remote from the inverter and the prevailing
environmental conditions aren't perfectly stable. One commonly overlooked
variable here is wind. We don't measure the windspeed during our process
and, frankly, without a long string of site-specific measurement data, we
couldn't factor it in anyway. What does matter is the effect of wind on
module temperature. A slight change in windspeed or direction can have a
very dramatic effect on module temperature and, by extension, power output
characteristics. If you record the module temperature and inverter displayed
output at slightly different times, you can end up with very different
results. Same with erratic irradiance. If you have fluctuating temperature
and irradiance, good luck trying to crunch the numbers to any significant
degree of accuracy.
 
Thirdly, you need to perform the procedure separately for each inverter
(system). Some will argue about this, but the fact of the matter is, you
either do them separately or you accept a much wider accuracy tolerance. If
your customer is of the belief that they will regularly see the DC PTC
rating of the array show up as AC Power Output on the inverter display....
Their glasses are probably too thick to accept wide tolerance ranges.
Lifetime thick-lenser typing this, BTW...
 
Fourthly, if that's a word. Instrumentation accuracy tolerances are what
they are. Some are published. Some are not. Some are lies, published as
facts. They are all "plus or minus something". Hard part is understanding
when it's plus and when it's minus and knowing what to do with that
information. Pyranometers are NOTORIOUS for range of accuracy issues, and
they ALMOST NEVER match their own specifications in the irradiance range you
are currently measuring. If you are relying on a weather-station or similar
DAS info to provide environmental data, you have some other issues to
contend with. Latency being the biggest underlying culprit. DAS systems are
wonderful tools for establishing trends and identifying out-of-character
performance over time. They suck for instantaneous evaluation. Suck is
actually an incredibly insufficient term for describing how bad they really
are for this application. Get handheld instruments and use them. If your
customer wants to do this, they need to spend $200. ($120 for General
Instruments DBTU-1300 Solar Power Meter and $80 for an Infrared Temperature
Probe. Best price I've found for the 1300 is here:
http://www.ambientweather.com/gedbtu1300.html)
 
Fifthly, measure irradiance at the Plane of Array! Any other methodology for
this exercise is bull$&!t. I don't give a rat's hind end what the
instructions say. If you've only got one pyranometer, throw the leveling
base away. I don't care that the bonehead at the monitoring company tells
you they have some algorithm that compensates horizontal irradiance to
system performance. It is totally OK to measure irradiance in a horizontal
plane when going thru this exercise. As long as you ignore that value and
also measure irradiance at the Plane of Array.
 
Sixth is understanding that baseline power measurements should be made with
zero shading on the array. Period. There is NO MEANINGFUL BENEFIT to going
thru this exercise if any portion of the array is shaded. NONE. When dealing
with shaded arrays, the only metric that is meaningful is the amount of
energy produced over time. And that's not what we're doing here.
 
Seventh is sort of a mixed-bag of best practice and potential to bite you in
the ass. In theory, the "best" time to evaluate the power output of a system
is when the sun is at solar noon relative the array azimuth. This isn't
always practical. Particularly with array azimuths closer to East or West
than South. In addition to actually having to be present when solar noon
rolls around, the weather also needs to cooperate by moving the clouds over
to the neighbor's house. Even if the timing is not a problem, during
non-winter weather, the module (cell) temperature is likely to be higher and
climbing at this time than others. Which does a couple things.... Generally
speaking, it drives the operating voltage lower when the potential
irradiance is at its peak for the day. When you go thru the exercise of
isolating various factors, you will find that module temperature has just as
dramatic an effect on system performance as irradiance does. When you really
dig into this, you will find that the published Coefficient of Temperature
for Power is pretty much erroneous at module temperatures above about 45C.
Actual observed outputs with high irradiance and high module temperatures
are generally lower than your theoretical "normalized" or "expected"
calculations. Which, of course, results in a phone call or email from your
customer. In practice, I basically "open up" the tolerances as the module
temperature rises.
 
Ocho. Soiling. Unless the modules are sparkling clean, there is going to be
some effect from soiling. Estimating actual soiling effect based on
observation alone requires experience. I'm not going to get into a class on
evaluating soiling effects in instantaneous performance verification here.
Suffice it to say that, if there is soiling at all, it's probably degrading
system output between one-quarter and ten percent. The same soiling will
have differing power characteristics at different temperatures and
irradiances throughout the day. 
 
Ninth is Age. How old is the system? Do non-module components degrade in
performance over time? Pick an annual degradation rate and use it. We used
to use 1% per year, but I think that practice is waning. I've seen a lot of
empirical monitoring data that indicates modern systems seem to be aging
without any meaningful degradation at all out to Year 5 or so. Nevertheless,
I recommend multiplying age in years by a factor of about 0.9925. This is
really rather arbitrary on my part... PV system output doesn't degrade from
100% on Day 365 to 99.25% on Day 366. In practice, I believe that we should
account for some aging and this is the balance I came up with. Pick your own
number if you like.
 
Tenthly should have been first. Shading. This procedure should NOT be done
if any portion of the array is shaded. Not by a layman. Not for the purpose
of evaluating system performance. None. Zero. Zip. Zilch. Nada. 
 
Eleventhly is a little moot at this stage, but worth mentioning. If you had
sold the customer a micro-inverter system, they would already know their
answer.
 
But let's move on to actually quantifying the factors and values in the
formula.
 
The Total STC Watts value should be self-explanatory in this forum. I will
reiterate that you need to do this on an inverter-by-inverter basis, so be
sure to only count the array that is connected to the inverter being
evaluated. This is a Constant value for each system.
 
The Irradiance Factor is pretty straightforward, too. To calculate this,
divide the measured Plane of Array Irradiance by 1000. eg. 850 W/M2 / 1000 =
0.85 Irradiance Factor. This is a Variable value that must be measured and
recorded each time you go thru this process.
 
The Temperature Factor is not equal to the temperature. This factor is used
to represent the difference in power between STC and the conditions under
which this test is performed. You need to measure the temperature, in
degrees Celsius, at the back of the module every time you do this test. (I
recommend using an infrared temperature probe and targeting the center of a
cell in the middle portion of a module. This can be difficult on low-profile
flush-mounted rooftop arrays. Do the best you can.) The SolarPro article
uses the Mfr temperature coefficient of power for this factor. In a
commissioning environment, this is the proper thing to do. In a "homeowner
wants to check his system every now and then" environment, not so good. For
CSi modules, I use a flat "half-percent per degree Celsius method". This
makes the math quite simple and straightforward. To compute the Temperature
Factor, subtract 25 from the measured module temperature to get your
temperature differential. Multiply that number by 0.005. Subtract this
number from 1 and you have your Temperature Factor. In practice, it goes
like this: A temperature differential of 25 degrees equals 12.5% loss or a
Temperature Factor of 0.875. A TD of 40 degrees equals a TF of 0.80.
 
And now we're into the System Derating Factor. Which combines Variable and
Constant values. In a commissioning environment, this may be acceptable. In
the homeowner case, not so good. I recommend stripping out the Variable
factors and treating them separately. In my own practice, I have always
treated these separately and strongly urge any and all to do the same.
Perhaps we should elaborate on the various values contained in the supposed
System Derating Factor.... Yeah. Good idea. As outlined in the SolarPro
article, these include 1) Wiring Losses, 2) Module Soiling Losses, 3) Module
Mismatch, 4) Module Nameplate Tolerance, 5) Shading, 6) Inverter Efficiency,
and 7) Age. I consider Soiling, Shading, and Age to be Variable values.
Wiring, Mismatch, Tolerance, and Efficiency losses are System Constants
specific to the system being evaluated. We can argue about whether or not
inverter efficiency is a constant or variable value, but I'm calling it a
constant and think you should too. I think it is more than fine to combine
these under the System Derating Factor heading. In order to combine
individual factors into a single composite factor, you multiply them by one
another.
 
A) Total wiring losses need to be accounted for and converted to a positive
factor format instead of % loss. If you don't do this correctly, you will
inadvertently end up with skewed numbers. I find that the best way to do it
is to think of this as a "wiring efficiency" factor instead of losses. If
you have a DC loss of 1.5%, that's actually 98.5% efficient or a Wiring
Efficiency Factor of 0.985. You treat the AC in the same manner. Multiply
your AC efficiency by your DC efficiency to get your system efficiency
factor. eg. A DC wiring loss of 1.5% & an AC wiring loss of 0.5% equate to a
System Wiring Efficiency factor of 0.980075, which I would round to 0.98.)
 
B) Module Mismatch needs to be calculated. My methodology is to base this on
the Power Tolerance Range. The Power Tolerance Range is the sum of the
absolute + and - values of the module Power Tolerance. A +10%/-5% module has
a 15% Power Tolerance Range. I multiply the PTR by 0.25 to get a Mismatch
Loss value. Subtract this number from 1 to get your Module Mismatch Factor.
eg. A +10%/-5% module would have a 3.75% Mismatch loss, which equals a
0.9625 Module Mismatch Factor. 
 
C) Power Tolerance Factor. This is pretty easy from my standpoint. Since we
are calculating a minimum acceptable value, I say use the Minus value of the
Power Tolerance and convert that to a factor. eg. A +10%/-5% module has a
0.95 Power Tolerance Factor.
 
D) Inverter Efficiency Factor. I will probably differ from some on this. My
methodology is to subtract 1% from the CEC Weighted Efficiency value. eg. An
inverter with a CEC weighted efficiency of 95.5% would have an Inverter
Efficiency Factor of 0.945.
 
Multiply A * B * C * D to get your System Derating Factor.
 
Age, Shading, Shading. Hmmm. What should we call this factor? 
 
I say it shouldn't be treated as a single factor, so no, it's not the ASS
factor. These are distinct and discrete variables that need to be treated
that way. As such, my recommended formula gets longer than the one mentioned
earlier on. It looks like this:
 
Minimum Expected Power Output = Total STC Watts * Irradiance Factor *
Temperature Factor * System Derating Factor * Age Factor * Shading Factor *
Soiling Factor.
 
Age is pretty straightforward, as discussed above. Figure out your Age
Factor and plug it in to the formula.
 
If you follow my methodology, the Shade Factor would be 1.0 since no testing
can be done if there is any shade at all on the array.
 
Soiling is a dirty beast. For instantaneous measurements, you do need to
come up with a Soiling Factor. And that's hard to describe to somebody. I
generally tell people to think of it this way: "Really, really dirty glass
is gonna max out at about 10 percent loss. This is very uncommon. Stuff like
caked mud, lots of bird doo-doo, restaurant grease covered in dirt. If you
can't clearly see all the individual cells, you have really, really dirty
glass. Kinda-dusty but mostly-clean is gonna ding ya about one or two
percent. If you can blow on it and all the dust blows clear, that's
kinda-dusty. Really-dusty is gonna be something like five or six percent.
Really-dusty won't just blow off. You need to brush it or wash it off, but
you can still clearly see the individual cells everywhere. If you never wash
your modules and just let the rain do it, in most cases the very worst
you'll see is about eight percent loss during part of the year. Washing
modules with hard water is one of the worst things you can do. So don't do
it." And I leave it at that. It truly is a judgement guess and it will vary
by site and season. Unless the array was just cleaned, I use a factor of
0.99 or lower.
 
The way to provide this to your customer is in a spreadsheet. Have the
constant values for their system filled in and fields for them to enter
environmental data and observed inverter output power. You can have the
sheet automatically calculate the expected output power and compare that to
the observed value. 
 
I suspect that, once you think all this thru, the CSI Field Verification
Form methodology will be more than sufficient. ;) Remember to tell your
customer that they have to at least by a handheld pyranometer if they want
to do instantaneous output verification.
 
$0.02001
 
Solar Janitor
 
  _____  

From: re-wrenches-bounces at lists.re-wrenches.org
[mailto:re-wrenches-bounces at lists.re-wrenches.org] On Behalf Of Marco
Mangelsdorf
Sent: Tuesday, October 12, 2010 6:17 PM
To: 'RE-wrenches'
Subject: [RE-wrenches] insolation v. actual output



Could someone please provide me with that generally accepted equation when
it comes to estimating AC output from a PV array versus the STC rating?

 

That is, I'm looking for that equation which estimates the losses due to mod
mismatch, soiling, wire losses, etc., etc.  I've got someone who mistakenly
expects their PV array to put out 90 or more percent of the STC rating.

 

Thanks,

marco

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20101013/bced29d2/attachment-0003.html>