System Overview

Technical Stuff

This page will describe the system in form of an overview - with details about the components

Components:

• Array

  • cells

    • this data is old. If you have newer data for me , please send me a link/URL. Thanks!

  • Material	Level of efficiency in % Lab	Level of efficiency in % Production
    Monocrystalline Silicon	approx. 24	14 to17
    Polycrystalline Silicon	approx. 18	13 to15
    Amorphous Silicon	approx. 13	5 to7

     

  • modules

    • The modules (panels) are Sharp 160Watt poly-crystalline panels, each on consisting of 6x8 = 48 cells in series

      • Voltage at max. power output: 22.8V

      • Current at max. power output: 7.02A

      • 7.02A x 22.8V = 160W

      • conversion efficiency (c.e.): 12.2% (module level - that means the system runs at approx. 11.5% overall efficiency (see "Losses", below).

        • (my personal favorite would have been Sanyo's HIT modules), but AES preferred the Sharp modules

        • Oregon has many low light days and can have hot summers. This combination bodes well for a-Si solar cells. However, if you don't have much (e.g. limited roof size), you're probably better off with Sanyo's HITs

    • The panels are approx. 4'5" x 3'3" in size (1.8" thick) and weigh about 35 pounds (this type of data you need if you want to figure out how much to stick onto your roof)

    • Another Voltage/Current pair you'll find in every spec is: Voc and Isc; open-circuit Voltage and short-circuit current. With the three reference points Voc, Isc and the max. power point you can describe the main characteristics of a module in one simple graph. It's called the I-V or U-I curve, and it looks just like the curve of a diode - surprise surprise - because that's what it basically what it is, a photo diode. A simple p/n junction.

    • If you think of cells in a module not as diodes, but as batteries in series, you can see how current and voltage correlate. This is also a good model, if you want to explain why a cell covered by a leaf has such a devastating effect: it turns into a producer (battery) into a consumer (resistor). Here's another site with a nice overview.)

  • frame

    •  

• Inverter (Fronius, PV Powered)

  • Displays: The Fronius Display is a reflective LCD (with LED front-light), while the PV Powered display is emissive (easier to read but consumes a few mW more on average).

  • data logging

    • Both allow remote monitoring but the PV Powered solution seems easier to install, requires less components and allows the user convenient access to the data via the Internet automatically.

      Fronius offers a nice piece of software which allows you to quickly download your power creation history, and lets you draw the data in hourly (down to 5 min intervals) , daily, monthly or in yearly format.
       

      • This screen shot is special for Western Portland, as it shows an absolutely sunny day in March. So you can see the maximum output curve in Spring: . It's a nearly perfect sine wave and shows the upper limit for each time of day.  Compare this to the graph from the day before, so this: to this: . (see next paragraph) You can see when there were clouds and approx. how thick they were, on the 18th.
         

    • The area underneath the curves represents the amount of Energy ([kilo-]Watts multiplied by hours =>  kWh). If you were to take the graph and change the sine wave into a square/rectangle with the highest point being 5,750W - the highest [theoretical] output to be expected from the array- you'd get the amount of sun hours on the horizontal axis. The *average* amount of sun-hours for the greater Portland (Oregon) area for the 19th of March is somewhere in the 1.5-2h range. Which means I could expect each of my arrays to put out about 5.750kW x 1.5h =  8.625 kWh per day on average. [Or 17kWh per day on average in March for my 11.5 kW system] Since 3/19 was so beautiful, the Fronius inverter actually registered 36.68kWh produced on that day. So 4x the average for this time of the year. That means my total system (11.5kW) produced almost 74kWh, today. This is encouraging, because according to my electricity bill my average energy consumption is 50kWh/day in March. So I became a net energy exporter, today. As a matter of fact, since I didn't have to heat the house, today (I have large windows facing south), I exported quite a bit of energy, today (about 40kWh).

      • I should note here, that although my system is rated as an 11.52 kW system, it probably will never reach that point, since all PV manufacturers quote their panels too high. Measurements of 1000W/m2 and 25degreesC module temp (PTC = PVUSA Test Conditions) or 20C ambient (STC = Standard Testing Conditions) seem to be extremely rare - as a matter of fact I don't think they ever happen in Portland. We'll see in June! See NREL's web pages for reference on STC and PTC.

      • Temperature Dependency of efficiency:
        Just out of morbid curiosity I overlaid the two graphs from March 18th (cloudy) and March 19th (sunny), and I noticed something interesting. : Unless the Fronius software shows an inconsistency - which I doubt -  the following takes place: On a cloudy day the output of the system is actually higher than on a sunny day (same time of day) for a brief period. This is probably caused by the fact that semiconductors don't like heat and the system is cooled by the rain (or during cloudy conditions), so that when sun comes out and the rays hit the panel, the lower temperature causes a higher conversion efficiency.

      • So just for fun, here's a trivia question:  Could an additional factor be that the clouds reflect some of the light, so the light bounces around (see picture), to increase the insolation (=irradiation) ? Maybe few permille?
        What do you think? Please let me know. Thanks
         

      • The Sharp specs didn't contain any reference, but I found from a competitor's product, that 3 degrees K of temperature rise cause the efficiency of a polycrystalline solar cell to go down by about 1%, which should explain most of the "overshoot" I found in the graphs, above. I need to make the experiment on a warm sunny day, sometime: although I don't have a good "thermometer" to measure the surface temperature of a solar cell precisely, I can at least pour water over the system to check on the effect of cooling the modules (Thanks to Marty for the idea).  The difference in output seemed to be in the 10-11% range, so if the output increases by 11% and we assume that only the temperature difference is the cause, we are looking at approx. 35 degrees K difference between the panel in the sunshine and in the rain.(Ambient temp. on that day was approx. 7-9 degrees C, by the way
        I wonder, however, just how much the effect of indirect (reflected, refracted) sunlight is. Any info you can point me to (link/URL) would be appreciated. Thanks.

     

• Interconnects

   

  • counter

    • One (analog) counter counts the total amount of energy coming out of the inverters; regardless of where that energy ends up being sent to.

    • PGE (Pacific Gas and Electric - my power company) installed another counter - a digital one that shows "energy bought" and "energy sold". What ticks me off about PGE's policy is that they made the counter indicators so convoluted and difficult to read that the user has a hard time figuring out when he's actually exporting energy. No clear indication of which way the current is flowing at the very moment. All people I asked (and who experienced the same frustration) were sure it's on purpose - meaning that PGE doesn't want the end user to know when he's feeding into the grid.
      
         

• Losses:

  • as the electrons go from generation to consumption, there are several places of loss (not a complete list!):

    • inside the cell is a p/n junction (B-doped Si ['missing' electrons] and P-doped Si [excess electrons]).  This p-n junction has 2 big "losses":

      • a) many of the light rays hitting the p/n junction don't have enough energy to kick an electron over the junction (or too much and that's not effective, either), so only a limited spectrum of the light hitting the cell is usable for the cell. One way to capture more wave lengths of light is to put 2 or 3 junctions in series (each with a different material composition). Tripple junction is just about the maximum we can sensibly do, today.

      • b) Electrons kicked out of the P-doped (n-side of the junction) into the B-doped region (p-side of the junction) sometimes don't go through the load (i.e. "not towards your house") but recombine immediately. Ooops. Wrong direction, electron! Part of the reason can be that the released electrons aren't captured soon enough (that's what these tiny little Aluminum lines are for: capturing loose electrons. However some electrons don't even make it that far - no matter how closely you put the Al lines together.

    •  Then there's the cell itself:

      • a) as mentioned above, the conductive lines/wires on the cell need to be rather close to each other - however, if they are too close together (and/or too thick), they block too much light, since they're usually on the top. Some companies managed to put the contact on the back. (this is an outdated graphic. wafer based solar cells are usually 200 to 250 um thick, if I remember correctly.

      • b) then there's the reflection of some of the light from the surface of the cell - one remedy, here, is to roughen up the surface to make it AR (Anti Reflective). AR films as used in the display industry seem to be a bit to expensive to use in solar cells, just yet, but I wouldn't be surprised to see them in the market, in the future.

      • c) another inefficiency factor comes from the fact that many solar cells (like all semiconductors) don't like heat. a-Si (amorphous Silicon) solar cells buck that particular trend a bit.

      • d) another factor that reduces conversion efficiency is the ageing of solar cells - it happens mostly in the first few months of exposure to - hold on to your hats: UV. A Mr. Staebler and a Mr. Wronski described this effect first, so it's called the Staebler-Wronski effect. a-Si is more affected by this than p-Si or c-Si (poly-crystalline or mono-crystalline Silicon) solar cells.

    • Connecting the cells together

      • ... is done with Al "wires" screen-printed onto the cell. Connections and conductive material have a finite ohmic resistance, so there's a minor source of loss, there.

      • If you have a low quality cell in a series it reduces the efficiency of the whole series.

    • Connecting the panels together to an array

      • most of the losses will be in the connectors between the modules/panels. Some resistance may be found in the termination in the junction box. However from a system perspective, these losses are minimal.

    • Connecting the arrays to the inverters

      • transporting several Amps of high Voltage DC across any distance will introduce losses.

      • in my particular system the estimate was that approx. 1% of the generated energy is lost in transport.

    • Converting from DC to AC

      • Inverters are usually in the mid-nineties (92-96%). Squeezing out just one percent more is an accomplishment for an inverter manufacturer. Think about it: If you have a 5200W inverter and you loose 5% in conversion, that means you are heating up the relatively small box with 250W. This is significant, because inverters don't convert electricity as efficiently when they're hot.  This is why inverters are often passively cooled and mounted in a space that has air-flow.

    • see http://rredc.nrel.gov/solar/codes_algs/PVWATTS//system.html for "Derate Factors" which gives you some idea of what other losses a system may face.

Other items I'll talk about: learnings, recommendations, bureaucracy, planning