Unisolar array design issue

My firm is rectifying a horrendous install (many skinned PV cables, wrinkled laminates etc) by a now defunct high profile installer of hybrid PV/thermal systems.

There are three strings of Uni-solar laminates on a Sunnyboy 4000. Two 9 strip strings of PVL136s (Vmp 33.0/Voc 42.2/105W) and one 12 strip string of PVL68s (Vmp 16.1/Voc 23.1/ 68W)

The tech support for the SMA inverter (Unisolar PV is belly up) said the imbalance in the strings would average out, this has got to come at a cost of efficiency.

My question:

Why would the designer use only 12 of the PVL68s? With a Vmp of the string currently at 193.2 and the PVL136 strings at 297, would it make sense to add an additional 2 PVL68s to bring that string to 230V reducing the imbalance, increase the the operating time (the SB4000 has a starting voltage of 260v and being asthetically better as the additional two strips would align with the strings above?

Comments

  • inetdog
    inetdog Solar Expert Posts: 3,123 ✭✭✭✭
    Re: Unisolar array design issue
    TechMan wrote: »
    My firm is rectifying a horrendous install (many skinned PV cables, wrinkled laminates etc) by a now defunct high profile installer of hybrid PV/thermal systems.

    There are three strings of Uni-solar laminates on a Sunnyboy 4000. Two 9 strip strings of PVL136s (Vmp 33.0/Voc 42.2/105W) and one 12 strip string of PVL68s (Vmp 16.1/Voc 23.1/ 68W)

    The tech support for the SMA inverter (Unisolar PV is belly up) said the imbalance in the strings would average out, this has got to come at a cost of efficiency.

    My question:

    Why would the designer use only 12 of the PVL68s? With a Vmp of the string currently at 193.2 and the PVL136 strings at 297, would it make sense to add an additional 2 PVL68s to bring that string to 230V reducing the imbalance, increase the the operating time (the SB4000 has a starting voltage of 260v and being asthetically better as the additional two strips would align with the strings above?

    Well, first of all the imbalance is well beyond the rule of thumb 5% difference in currrent (series) or voltage (parallel), so it deserves a more detailed analysis.

    And that involves more than just comparing Voc to Voc and Vmp to Vmp.

    Your two strings have a Voc of 379.8 and, as you noted a Vmp of 297 while the third string has a Voc of 277.2 and a Vmp of 193.

    What is the result of this when the MPPT controller looks at them?

    If the controller starts at zero current and high voltage, it will start at 379.8 and start to draw more and more current, reducing the voltage, and getting a steadily increasing amount of power until it hits the MPP of the strings of 12 at 297 volts. The third string, with its Voc of 277 has not even started to deliver any current yet!
    As the controller continues to lower the voltage and get slightly more current, it will see that the total power is decreasing.

    Unless it continues to scan below 277 volts, it will not even know that the third string exists. And even if it scans lower, it may well find that by the time it starts getting current from string 3, the effect of the lower voltage on strings 1 and 2 will cause a lower total power at the next point where the power stops rising and starts falling again (somewhere above the Vmp of the third string at 193.2) than it got without string 3 at all!

    If you added two panels to make a string of 14 with Voc of 309 and a Vmp of 230, then there is still a great enough difference between the two Vmp values that the sum of the two power outputs will be a lot less than the sum of the two maximum power outputs. Since there are two strings with Vmp of 297, the absoute maximum power point still may not be getting much of a contribution from string 3.

    If you are able to add 4 panels to the third string instead, bringing its Vmp up to 262, then you would have a match within 5% and you would get very close to the sum of the maximum power of the three strings.

    The only way the current configuration might be viable would be if the third string is also pointing in a different direction and delivers its max power at a different time than strings 1 and 2. And even then, to be sure that the controller will make the right decision, you still need to get string 3's Voc above the Vmp of the first two strings.
    SMA SB 3000, old BP panels.
  • niel
    niel Solar Expert Posts: 10,300 ✭✭✭✭
    Re: Unisolar array design issue

    if the cash is available take out the unisolar laminents and put in some real proper pvs with more power that takes up less room and mount them so they can be properly pointed more toward the sun.
  • TechMan
    TechMan Registered Users Posts: 5
    Re: Unisolar array design issue
    niel wrote: »
    if the cash is available take out the unisolar laminents and put in some real proper pvs with more power that takes up less room and mount them so they can be properly pointed more toward the sun.


    Unfortunately the PV is part of a hybrid system with the thermal in back of the standing seam. I wanted to use thermal tubes from the on-set, but alas was not in control of these items at the time.

    The roof is at a 74 degree tilt and pointed very close to solar south.
  • TechMan
    TechMan Registered Users Posts: 5
    Re: Unisolar array design issue
    inetdog wrote: »
    Well, first of all the imbalance is well beyond the rule of thumb 5% difference in currrent (series) or voltage (parallel), so it deserves a more detailed analysis.

    And that involves more than just comparing Voc to Voc and Vmp to Vmp.

    Your two strings have a Voc of 379.8 and, as you noted a Vmp of 297 while the third string has a Voc of 277.2 and a Vmp of 193.

    What is the result of this when the MPPT controller looks at them?

    If the controller starts at zero current and high voltage, it will start at 379.8 and start to draw more and more current, reducing the voltage, and getting a steadily increasing amount of power until it hits the MPP of the strings of 12 at 297 volts. The third string, with its Voc of 277 has not even started to deliver any current yet!
    As the controller continues to lower the voltage and get slightly more current, it will see that the total power is decreasing.

    Unless it continues to scan below 277 volts, it will not even know that the third string exists. And even if it scans lower, it may well find that by the time it starts getting current from string 3, the effect of the lower voltage on strings 1 and 2 will cause a lower total power at the next point where the power stops rising and starts falling again (somewhere above the Vmp of the third string at 193.2) than it got without string 3 at all!

    If you added two panels to make a string of 14 with Voc of 309 and a Vmp of 230, then there is still a great enough difference between the two Vmp values that the sum of the two power outputs will be a lot less than the sum of the two maximum power outputs. Since there are two strings with Vmp of 297, the absoute maximum power point still may not be getting much of a contribution from string 3.

    If you are able to add 4 panels to the third string instead, bringing its Vmp up to 262, then you would have a match within 5% and you would get very close to the sum of the maximum power of the three strings.

    The only way the current configuration might be viable would be if the third string is also pointing in a different direction and delivers its max power at a different time than strings 1 and 2. And even then, to be sure that the controller will make the right decision, you still need to get string 3's Voc above the Vmp of the first two strings.

    Thanks INETDOG.

    Kind of what I expected. Trying to remedy this has been a challenge to say the least. I am going to site tomorrow to see if I can add 4 total to the lower string. May have the room to one side of the upper strings. SMA tech said the turn on for the SB4000 is at 260 volt consistant with your analysis. Unfortunately, the installing company has provided no design data (through obvious lack of diligence it would probably be useless) and I have yet to find any design simulation software to accomodate this scenario. I would like to simulate removing a panel from each of the upper strings and adding the two the bottom.

    In the long term, we may approach other previous installs to rectify deficient or non-functional systems. The previous installer was in business for 8 or 9 years, so there should be a string of these.
  • ggunn
    ggunn Solar Expert Posts: 1,973 ✭✭✭
    Re: Unisolar array design issue
    TechMan wrote: »

    The roof is at a 74 degree tilt and pointed very close to solar south.
    74 degrees? Ye gods, how do you work on a nearly vertical surface? Or do you mean that it's 74 degrees off vertical, which would be a tilt of 16 degrees?
  • TechMan
    TechMan Registered Users Posts: 5
    Re: Unisolar array design issue
    ggunn wrote: »
    74 degrees? Ye gods, how do you work on a nearly vertical surface? Or do you mean that it's 74 degrees off vertical, which would be a tilt of 16 degrees?

    A very contemporary home. Peak on low side is 42', on high side is 35'. Had to rent a 60' Genie Lift to get to top. With the slope, you pull the man basket off the wall about five inches and you hardly have to lean to get at the connections. Becoming costly though, $2600 just for the lift, but other wise the install is useless.

    I did find some PVL 68s on Ebay to lower the imbalance. Fortunately, interior is not yet finished, so resolving the pinched wire under the standing seam can be done via a utilty chase.

    Will the fun ever end?:D
  • TechMan
    TechMan Registered Users Posts: 5
    Re: Unisolar array design issue
    inetdog wrote: »
    Well, first of all the imbalance is well beyond the rule of thumb 5% difference in currrent (series) or voltage (parallel), so it deserves a more detailed analysis.

    And that involves more than just comparing Voc to Voc and Vmp to Vmp.

    Your two strings have a Voc of 379.8 and, as you noted a Vmp of 297 while the third string has a Voc of 277.2 and a Vmp of 193.

    What is the result of this when the MPPT controller looks at them?

    If the controller starts at zero current and high voltage, it will start at 379.8 and start to draw more and more current, reducing the voltage, and getting a steadily increasing amount of power until it hits the MPP of the strings of 12 at 297 volts. The third string, with its Voc of 277 has not even started to deliver any current yet!
    As the controller continues to lower the voltage and get slightly more current, it will see that the total power is decreasing.

    Unless it continues to scan below 277 volts, it will not even know that the third string exists. And even if it scans lower, it may well find that by the time it starts getting current from string 3, the effect of the lower voltage on strings 1 and 2 will cause a lower total power at the next point where the power stops rising and starts falling again (somewhere above the Vmp of the third string at 193.2) than it got without string 3 at all!

    If you added two panels to make a string of 14 with Voc of 309 and a Vmp of 230, then there is still a great enough difference between the two Vmp values that the sum of the two power outputs will be a lot less than the sum of the two maximum power outputs. Since there are two strings with Vmp of 297, the absoute maximum power point still may not be getting much of a contribution from string 3.

    If you are able to add 4 panels to the third string instead, bringing its Vmp up to 262, then you would have a match within 5% and you would get very close to the sum of the maximum power of the three strings.

    The only way the current configuration might be viable would be if the third string is also pointing in a different direction and delivers its max power at a different time than strings 1 and 2. And even then, to be sure that the controller will make the right decision, you still need to get string 3's Voc above the Vmp of the first two strings.

    Thanks again for the great insight.

    What about pulling a panel out of each of the uppers and adding two to the lower? Brings Vmp of uppers down to 267 and the Voc of the lower up to 323 and separates the Vmp by about 15%. Does the loss of 210w of strip outweigh the gain of the closer balance?