Solar Panels and Shedding Snow

13»

Comments

  • mtdoc
    mtdoc Solar Expert Posts: 600 ✭✭
    Re: Solar Panels and Shedding Snow

    Since Solar Guppy has already done this real world experiment and achieved the results I've explained here there's no reason to repeat it.

    Hmm. Kind of like those cold fusion experiments - no reason to repeat them???

    Sorry, but no single experiment really proves anything ... Any conclusion that cannot be verified by repeat experimentation is .....

    Inetdog makes a good case for why in theory there is not much to be gained from a second cc in Chris's case. In practice it may be different but the only way to determine this is by testing and no, one person's report does not convince me. It would be good to get some confirmation. :-)
  • BB.
    BB. Super Moderators, Administrators Posts: 33,422 admin
    Re: Solar Panels and Shedding Snow

    It is not that the "lower Vmp-array" set of panels "draws power"--It is the issue that there are (at least) two different power peaks. Standard MPPT controllers have not been designed to find the maximum of two different peaks (not to say that they cannot be designed this way--just is not not been done in the past).

    Personally, I would have thought that having two different array orientations and two different Vmp-arrays would not have been that great of difference (the colder array is simply going to have significantly less light available--so if it is "ignored" in the MPPT search, it is not a huge loss). However, Solar Guppy (a few years ago) was very addimate that this was not the case and it was important to have two different MPPT units for the East/West Arrays.

    Is this still true today and with every controller out there--I do not know.

    boB from Midnite has not commented (that I recall) on the issues between an East vs West array--But it would be interesting if he does.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Solar Panels and Shedding Snow

    Solar Guppy tests solar equipment for a living. I imagine he's done this more than once and is pretty accurate about doing it right. If anyone wants to repeat the experiment, go ahead: it's your time and money to waste.

    Chris Olson lives in the frozen North like I do; panel Voltage in Winter can be 1.3 times what it is in Summer. That extra Voltage on an MPPT controller equates to output power. Having one cold array in full sun and one in full shadow (if we return to the notion of an East array and a West array both hung vertical) the Voltage difference between the two will be quite significant, especially on 3500 Watts of panel. Even using the ultra-conservative 10% figure that amounts to around 1kW hour per day, which is not to be dismissed.

    If this were two small arrays or a slight angular difference or a tropic location the difference would indeed be negligible. But it isn't so it isn't.

    As always, no one has to agree with me or follow my advice. It isn't my money being spent.
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Solar Panels and Shedding Snow

    Just so we are clear, the difference between the conditions of the east and west arrays that I was concerned about is that of insolation, not temperature, although there may be a temperature difference because of the difference in insolation which would complicate things more. I'm pretty sure that the Vmp that inetdog was talking about is that imposed by the MPPT circuitry, not the value on the data sheet.
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Solar Panels and Shedding Snow
    ggunn wrote: »
    Just so we are clear, the difference between the conditions of the east and west arrays that I was concerned about is that of insolation, not temperature, although there may be a temperature difference because of the difference in insolation which would complicate things more. I'm pretty sure that the Vmp that inetdog was talking about is that imposed by the MPPT circuitry, not the value on the data sheet.

    You are almost exactly right, and thanks for mentioning that.

    For any given panel, the rated Vmp will be the single Vmp value which corresponds to the test conditions of 1000w/m2, etc. There will also be a single value for Imp corresponding to that light input.

    The value on the graphs is the actual measured Vmp for a particular different level of light input. The MPPT controller will find that MPP for the panel array in the current light and shading conditions. That number will be a function of Irradiance, so maybe I should call it Vmp(Ir) instead of just Vmp. And when I say Irradiance, I am assuming that number factors in the angle of incidence, diffusion of light, and all those purely incoming-light-related factors. It should actually be Vmp(Ir,T) where T is the temperature, but for the moment we will ignore that. :-)
    My observation from the graphs is that, at least over the the range from 1/10 power to full power, Vmp(Ir) is never more than 5% lower than the nameplate Vmp. Imp(Ir) is roughly directly proportional to Ir.

    Finally, although the Vmp(Ir) is a definite function characteristic of the panel, the operating point of the CC may be something hopefully identical to but at least close to that Vmp(Ir) number as long as only one subarray is connected.
    Given that, the VCC (chosen by the controller's MPPT circuitry) will be very close to Vmp(Ir) of the array which is producing the most power. The rest follows from there, given that the other array will see that same VCC which will then be within 5% of the actual Vmp(Ir) of that other subarray.

    I may have lost the algebra-challenged posters, and not been formal enough for the mathematicians, but this is the best I can do to explain my belief in one post.

    One last comment: Although Vmp does not vary much with Ir and varies predictably and by less than 25% with T, shading (S) of part or all of any panels in the string is likely to reduce Vmp(Ir,T,S) by as much as the nameplate Vmp of one entire panel. In that case my 5%-based reasoning that one MPPT input will be good enough goes out the window.
    SMA SB 3000, old BP panels.
  • mtdoc
    mtdoc Solar Expert Posts: 600 ✭✭
    Re: Solar Panels and Shedding Snow
    Solar Guppy tests solar equipment for a living. I imagine he's done this more than once and is pretty accurate about doing it right. If anyone wants to repeat the experiment, go ahead: it's your time and money to waste.

    It would be great to see exactly what the test conditions that SG used were. Identical panels, shading, temperature differences , etc? A hot day on a Florida rooftop would likely produce dramatic differences between an array in direct sunlight and one that wasn't, wouldn't it? Much different conditions that the great frozen north where Chris resides.

    Perhaps SG is lurking and can help clarify his test conditions? I know his website is shut down. Does he still do testing for Xantrex? Was his testing of this issue done for Xantrex? I suppose Xantrex, selling MPPT controllers might not be considered an impartial party in this issue. Would be surprised if they told someone they do not need to buy another CC.....
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Solar Panels and Shedding Snow
    Voc doesn't actually exist on panels that are connected to a controller: no "open circuit" hence no "Voltage open circuit".

    My but this is getting complicated.
    [lecture mode]
    Voc (or Voc(Ir,T,S) exists as a property of the panel whether you can measure it at any particular time or not. By definition, it is the voltage which would be present when no current is flowing (which could be the result of an open circuit or just a connection to something else which is also at Voc.)
    If an MPPT controller does a scan all the way to the point where zero current if flowing from a single string, it will be able to read Voc and do whatever it wants to do based on that. If the scan passes through the Voc of a string with a smaller number of panels which is connected in parallel with the first string, there will be a point at which the CC is getting current input, but none of it is coming from one of the two strings. That voltage is the Voc of that string, even though it is not actually open circuited.

    If a lower voltage string (lv) is connected to a higher voltage string (hv) in parallel, there may be a point at which no current is flowing through the lv string in either direction while current is being drawn from the other, hv, string. And yes, if the output of the hv string is above the Voc value of the lv string, then current will flow in the reverse direction and the lv string will be consuming power. And this will be a very difficult situation for an MPPT controller to deal with. Especially if it has programmed into it an expectation of the shape of the I vs. V curve of the source.

    However(!) if the two strings are made up of identical panels, then there is no way, regardless of light condition (even including partial shading), in which the current Voc of either string can be less than the Vmp of the other string.
    [/lecture mode]
    Maybe someday we will reach a consensus MPPT analysis of a wide range of different parallel string, array/subarray, mismatched panels, etc., configurations. When that happens, it will deserve to be a prominent sticky thread, since I have not seen a complete analysis of this anywhere else yet.
    BTW, I would love to see the Solar Guppy results if somebody has a saved copy.
    SMA SB 3000, old BP panels.
  • mtdoc
    mtdoc Solar Expert Posts: 600 ✭✭
    Re: Solar Panels and Shedding Snow
    inetdog wrote: »
    If an MPPT controller does a scan all the way to the point where zero current if flowing from a single string, it will be able to read Voc and do whatever it wants to do based on that.

    Which explains why my Midnite Classics display the VOC for my PV strings.:cool:
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Solar Panels and Shedding Snow
    mtdoc wrote: »
    Which explains why my Midnite Classics display the VOC for my PV strings.:cool:

    This is probably a big point to the whole issue: the testing Solar Guppy did (and the controllers I'm used to) would be of the "old style" that did periodic sweeps, rather than the "new style" like the MidNite that readjust continuously. That may not sound like much, but in terms of coming up with the best power point it is easier to 'fool' the old units (conditions at the moment of sweep set the power point until the next sweep).

    If I had the money I'd buy one of those Classics and try the same tortures I've done to the MX60 on it. :p
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Solar Panels and Shedding Snow
    inetdog wrote: »
    You are almost exactly right, and thanks for mentioning that.

    For any given panel, the rated Vmp will be the single Vmp value which corresponds to the test conditions of 1000w/m2, etc. There will also be a single value for Imp corresponding to that light input.

    The value on the graphs is the actual measured Vmp for a particular different level of light input. The MPPT controller will find that MPP for the panel array in the current light and shading conditions. That number will be a function of Irradiance, so maybe I should call it Vmp(Ir) instead of just Vmp. And when I say Irradiance, I am assuming that number factors in the angle of incidence, diffusion of light, and all those purely incoming-light-related factors. It should actually be Vmp(Ir,T) where T is the temperature, but for the moment we will ignore that. :-)
    My observation from the graphs is that, at least over the the range from 1/10 power to full power, Vmp(Ir) is never more than 5% lower than the nameplate Vmp. Imp(Ir) is roughly directly proportional to Ir.

    Finally, although the Vmp(Ir) is a definite function characteristic of the panel, the operating point of the CC may be something hopefully identical to but at least close to that Vmp(Ir) number as long as only one subarray is connected.
    Given that, the VCC (chosen by the controller's MPPT circuitry) will be very close to Vmp(Ir) of the array which is producing the most power. The rest follows from there, given that the other array will see that same VCC which will then be within 5% of the actual Vmp(Ir) of that other subarray.

    I may have lost the algebra-challenged posters, and not been formal enough for the mathematicians, but this is the best I can do to explain my belief in one post.

    One last comment: Although Vmp does not vary much with Ir and varies predictably and by less than 25% with T, shading (S) of part or all of any panels in the string is likely to reduce Vmp(Ir,T,S) by as much as the nameplate Vmp of one entire panel. In that case my 5%-based reasoning that one MPPT input will be good enough goes out the window.
    Then why have MPPT at all on a system where shading is not an issue? According to a poster on the "other forum", if it were not for MPPT, inverters could act as voltage sources and we wouldn't need batteries to run off-grid.

    :D
  • Cariboocoot
    Cariboocoot Banned Posts: 17,615 ✭✭✭
    Re: Solar Panels and Shedding Snow
    ggunn wrote: »
    Then why have MPPT at all on a system where shading is not an issue? According to a poster on the "other forum", if it were not for MPPT, inverters could act as voltage sources and we wouldn't need batteries to run off-grid.

    :D

    Inverters as Voltage sources? That would be a neat trick. :roll:
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Solar Panels and Shedding Snow
    ggunn wrote: »
    Then why have MPPT at all on a system where shading is not an issue? According to a poster on the "other forum", if it were not for MPPT, inverters could act as voltage sources and we wouldn't need batteries to run off-grid.

    :D

    I will split that up into two parts.

    1. The "voltage source" comment is rather inane and and technically flawed. I do not support it in any way. Although the Vmp of the panel might be relatively constant over the solar day, that does not make the panel a voltage source. It will only deliver a strictly limited current at that voltage, which makes the panel a poorly regulated current source rather than a poorly regulated voltage source. And the comment about running without batteries is also totally absurd unless there is multi-cycle energy storage to take care of the time varying power demands of an AC waveform AND the load is constantly adjusted to match the output of the panels. Not going to happen except in certain very limited situations. The new SunnyBoy GTI with grid failure backup limits the maximum output to a fraction of the rating of the inverter and the panels and makes use of the energy storage inherent in its design to cover the AC waveform requirements. I think that if you look at the DC input of a conventional battery-driven inverter, you will see that the input current is nowhere near constant but varies at a 120 Hz frequency. This is especially true of MSW inverter designs. Any direct-from-panel inverter without energy storage will be limited to the AC output for which the PEAK DC current is at or below the current capability of the panels. (I spent more time on that than I intended, since I got caught up in points that I think have been under-addressed elsewhere.)

    2. The DC-to-DC conversion inherent in a MPPT device allows for better performance when the Vmp of the source is varying. For a fixed defined panel setup (regardless of orientation) at a fixed temperature, you could build a non-tracking DC-to-DC converter which would be maximally efficient. This is, I think, without dispute. But given that you must include a DC-to-DC converter (or the equivalent in the case of a GTI) you would either have to provide a manual adjustment for any particular string configuration and panel type or produce a device that only works well with one single panel configuration. For a small extra cost (given the flexibility of modern digital controllers) you can build a box that will accept a wide range of different arrays with no manual adjustment by the user. When you throw in the fact that even a fixed design array will show temperature effects and that for some home users some shading is inevitable, there is no economic reason not to go all the way to MPPT.

    Interesting additional background, from the Holt forum, credit to "Garrison":
    Apparently SMA no longer discourages this practice [paralleling strings with different orientation] with their inverters. The article linked below suggests that since peak string performance will be at different times, the inverter size can be smaller. From an economic perspective this can increase your rate of return for a given array size, which can make up for some of the power loss [they say].

    http://www.smainverted.com/files/201...-TEN122510.pdf

    A salient excerpt from that article (a bit hard to read in places because of its German-tinged grammar) is the following (emphasis mine):
    As an alternative to using a multi-string inverter, it is also feasible to operate both PV
    arrays on just one MPP tracker. As earlier investigations on existing PV plants with
    east/west-aligned generators have shown, in certain cases polystring operation only
    displays minimal mismatching losses, with the result that yield losses are significantly
    lower than one percent [1, 3]. The currents from the PV modules may vary
    significantly in the substrings over the course of the day, as the module surfaces are
    subject to varying intensities of irradiation. However, the voltage at the MPP is
    practically identical.
    The precondition for polystring operation is that the substrings
    have an identical arrangement, i.e., they have the same number of PV modules of
    the same type.
    Simulations we have performed confirm the results of the above investigations: With
    east/west alignment of the substrings, a yield loss of 0.25% is found in polystring
    operation as against multi-string operation.

    I think that the economy of having only one Inverter or CC with a single MPPT input more than compensates for a .25% loss in energy collection.
    As secondary benefit of using one device rather than two independent devices (not considering for the moment one device with two trackers) is that you can collect the full energy of a pair of E and W subarrays of 3KW each using only a 3KW device. If you used separate devices, you would need two 3KW devices to handle the same pair of arrays.
    SMA SB 3000, old BP panels.
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Solar Panels and Shedding Snow
    inetdog wrote: »
    I will split that up into two parts.

    1. The "voltage source" comment is rather inane and and technically flawed. I do not support it in any way. Although the Vmp of the panel might be relatively constant over the solar day, that does not make the panel a voltage source. It will only deliver a strictly limited current at that voltage, which makes the panel a poorly regulated current source rather than a poorly regulated voltage source. And the comment about running without batteries is also totally absurd unless there is multi-cycle energy storage to take care of the time varying power demands of an AC waveform AND the load is constantly adjusted to match the output of the panels. Not going to happen except in certain very limited situations. The new SunnyBoy GTI with grid failure backup limits the maximum output to a fraction of the rating of the inverter and the panels and makes use of the energy storage inherent in its design to cover the AC waveform requirements. I think that if you look at the DC input of a conventional battery-driven inverter, you will see that the input current is nowhere near constant but varies at a 120 Hz frequency. This is especially true of MSW inverter designs. Any direct-from-panel inverter without energy storage will be limited to the AC output for which the PEAK DC current is at or below the current capability of the panels. (I spent more time on that than I intended, since I got caught up in points that I think have been under-addressed elsewhere.)
    You do remember that the inane comment wasn't mine, don't you? Just checking...
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Solar Panels and Shedding Snow
    ggunn wrote: »
    You do remember that the inane comment wasn't mine, don't you? Just checking...
    Absolutely! Just joining you in bashing that comment, while replying to your question about why MPPT in one post. Hence the two parts. :-)
    SMA SB 3000, old BP panels.
  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Solar Panels and Shedding Snow
    inetdog wrote: »
    Interesting additional background,

    Continued...

    Another very helpful (although from a particular manufacturer and therefore some what biased :-)) is this White Paper from Fronius:

    http://www.fronius.com/cps/rde/xbcr/SID-05FAD872-5CCBCDF5/fronius_usa/PDF_fronius_ig_reaction_to_non_optimal_conditions_207252_snapshot.pdf
    The meat of the data comes in the form of a set of graphs of a series string with exactly one panel shaded at various levels.
    The point that I draw from this is that depending on the degree of shading, the best-solution MPP for the string will be near the Vmp of the unshaded string if the single panel is lightly shaded, while for a heavily shaded panel, the Vmp will be that for a shorter string which did not have the shaded module in it.

    So since the shapes of the two curves (heavily shaded versus unshaded) are so different, the MPP of the combination may end up being at a much lower V than for the unshaded string alone. This same effect is discussed further down the paper where they discuss mismatched numbers of panels in a string.

    In Figure 1, you see a fundamental change in the shape of the curve somewhere between 25% shading and 75% shading.
    In the 25% case there are still two local maxima, and the higher voltage one is the absolute maximum.
    In the 75% case there are two local maxima, but the lower voltage one is the absolute maximum.
    (I am looking at the slope of the I/V curve to determine where an area-under-the-curve maximum will occur, since they do not draw a power versus voltage curve.)
    SMA SB 3000, old BP panels.