Grid Tied AC coupling with Islanding Inverter

I am adding a control system to my Grid Tied (GT) PV system to allow better control of power generation versus power consumption. The goal is to minimize the cycling of the batteries while having the GT inverters AC coupled to an Islanding inverter and grid power is not available. The Islanding inverter and batteries will be purchased after the basic control system is working.

My PV system consists of 44 - M215 Enphase micro inverters with LPC244SM Samsung panels. The inverters are organized in 11 strings of 4 inverters. So each string will generate about 860Watts. The 11 strings come together in an AC combiner panel in the garage, then go to the PV meter, AC disconnect and into a 50amp feedback breaker in the Service Entrance. The system is being commissioned shortly.

The general plan is to measure the current from the PV system & to/from the Islanding inverter then adjust a variable diversion load and the number of strings on line to minimize the cycling of the battery while off grid. This should make the batteries last longer and possibly allow the use of a smaller battery.

So far I purchased 22 Solid State Relays (SSR) 6 Ratiometric Hall Effect sensors and an Arduino Uno micro controller. Still waiting for delivery. In addition I have Brultech home energy monitoring system that uses Current Transformers (CT). I may use some CTs with the Arduino if I can work out the details.

Initially I was planning to use a XW6048, but with new stuff coming out more frequently, I will keep my options open.

Comments

  • RCinFLA
    RCinFLA Solar Expert Posts: 1,484 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter

    Your use of Islanding inverter terminology is strange.

    A grid interactive hybrid inverter will not island. This is required in UL1741 spec. It refers to an grid tied / interactive inverter pushing power into the grid when grid is open. Which is a no-no.

    You are going to have better describe your variable diversion load. Since you mention 11 AC feeds and 22 solid state relays, along with hall effect sensor I assume you are intending to electronically select how many of the 11 GT strings are connected to output of hybrid inverter. You should check on how the SSR react to the directionality and leakage current from the GT array. My concern would be the the SSR don't cut off when you control them to.

    Also, the GT microinverter will take several minutes to reconnect. This may be problematic in tracking PV variable illumination changes and may drive your Arduino Uno micro controller crazy. The control action may be old info and inappropriate when the GT inverter finally come back online.

    At maximum illumination power you will be generating over 130 watts of heat in the 22 SSR's. 11 DPDT 5 amp contact rated relays, using normally closed contacts may be a better approach.
  • RK_Solar_Hopeful
    RK_Solar_Hopeful Solar Expert Posts: 69 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter
    RCinFLA wrote: »
    Your use of Islanding inverter terminology is strange.
    I am looking to keep the GT inverters producing if grid power goes down. An Islanding inverter similar to the XW6048 should produce the needed 240VAC, 60Hz the GT inverters need for syncing up.
    You are going to have better describe your variable diversion load.
    I think a straight resistive load like a heating coil or space heater would be the easiest to control. At this point I am making assumptions that need to be verified. I think a 1.5kW to 3kW diversion load should be large enough. Plus the heating element load can be controlled with a SSR that can be controlled by the Arduino. If the SSR is cycled on/off at a 50% duty cycle the heating element will give a 50% load. An 8% duty cycle gives 8% load. Very granular control.
    Since you mention 11 AC feeds and 22 solid state relays, along with hall effect sensor I assume you are intending to electronically select how many of the 11 GT strings are connected to output of hybrid inverter. You should check on how the SSR react to the directionality and leakage current from the GT array. My concern would be the the SSR don't cut off when you control them to.
    That is what I hope to do. The data sheets show the SSRs should work but that is another assumption that does need testing. The SSR data sheet is attached.
    Also, the GT microinverter will take several minutes to reconnect. This may be problematic in tracking PV variable illumination changes and may drive your Arduino Uno micro controller crazy. The control action may be old info and inappropriate when the GT inverter finally come back online.
    I am planning to make the diversion load large enough to cover intermittent load starts or minor illumination changes. For example; the A/C starts, the Arduino would see it and "start" enough PV strings to cover the load, at the same time the diversion load would be reduced to minimize the XW6048 battery usage, after 5 minutes when the PV strings sync up, the Arduino would see that and bring the diversion load back up to minimize the battery charging. During the opposite situation, the Arduino would give an immediate shutdown signal to the SSRs and response time is variable from instant to a 1/2 cycle of 60Hz (about 8.5ms). That is due to the "zero-crossing" nature of these SSRs. Needs to be verified to work.
    At maximum illumination power you will be generating over 130 watts of heat in the 22 SSR's. 11 DPDT 5 amp contact rated relays, using normally closed contacts may be a better approach.
    That is something I had overlooked! I was thinking about the ease of controlling the SSRs as a switch and missed the power consumption aspect of the SSR being a semi-conductor. The DPDT relays might be better. Another round of research...:roll: But this is why I brought it to the forum, Thank you!
  • RCinFLA
    RCinFLA Solar Expert Posts: 1,484 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter

    If this is going to be permitted and inspected you might want to install it stock and perform your 'customizations' after inspection approval.

    44 two hundred watt panels are pretty obvious to any power company personnel, county inspector, insurance survey, or disgruntled neighbors, that might cause you problems in the future. I would not recommend you do it without permit.

    Also, if you momentarily hit the XW6048 with more then 6kW during off grid backfeed it will immediately shut down. There is no battery current regulation on XW for backfeed power during off grid condition.
  • RK_Solar_Hopeful
    RK_Solar_Hopeful Solar Expert Posts: 69 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter
    RCinFLA wrote: »
    If this is going to be permitted and inspected you might want to install it stock and perform your 'customizations' after inspection approval.

    44 two hundred watt panels are pretty obvious to any power company personnel, county inspector, insurance survey, or disgruntled neighbors, that might cause you problems in the future. I would not recommend you do it without permit.
    The system has passed the city inspection and elec co inspection. Tomorrow the elec co is supposed to commission the system.:D I am absolutely getting the system properly turned on before I do anything.
    Also, if you momentarily hit the XW6048 with more then 6kW during off grid backfeed it will immediately shut down. There is no battery current regulation on XW for backfeed power during off grid condition.
    I was aware of the lack of current regulation of the XW6048. The immediate S/D if more than 6kW backfeeds is something I did not know.

    One of the project goals is to manage the Islanding inverter power to a level that keeps the battery happy. Not too much draw, not too much feed.
  • RCinFLA
    RCinFLA Solar Expert Posts: 1,484 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter

    The inverter is bidirectional. It is also maximum 6kW power. If you push more then that at it, the inverter will go into overload trying to push it to batteries.

    It may take a momentary overload. The question will be for how long and can your system react fast enough to reduce backfeed power push to prevent inverter from shutting itself down to protect itself from damage.

    Be aware that this backfeed power (when off grid) will have to be absorbed by batteries (or in combination with AC out load) so the battery bank must be large enough to take 100 amps or more of charging push at maximum inverter power capability. Depending on PV power your AC loads may be keeping backfeed power within 6kW limit but then someone switches off an ACout load so now the backfeed exceeds 6 kW.

    When on grid, the backfeed can go higher, up to the input AC breaker levels. (60 amps max) But you have to think about if grid suddenly drops and inverter-batteries might have to suck it up.
  • RK_Solar_Hopeful
    RK_Solar_Hopeful Solar Expert Posts: 69 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter
    RCinFLA wrote: »
    ...It may take a momentary overload. The question will be for how long and can your system react fast enough to reduce backfeed power push to prevent inverter from shutting itself down to protect itself from damage.

    Be aware that this backfeed power (when off grid) will have to be absorbed by batteries (or in combination with AC out load) so the battery bank must be large enough to take 100 amps or more of charging push at maximum inverter power capability. Depending on PV power your AC loads may be keeping backfeed power within 6kW limit but then someone switches off an ACout load so now the backfeed exceeds 6 kW.

    When on grid, the backfeed can go higher, up to the input AC breaker levels. (60 amps max) But you have to think about if grid suddenly drops and inverter-batteries might have to suck it up.

    The Arduino runs at 16mHz. From my reading it should be fast enough using a combination of fast response diversion load and SSRs/Relays. But I won't know until I've tried.

    If and this is a big IF the control system works well, then the backfeed to the battery bank would be limited to what it takes to recharge the battery bank at a reasonable rate.

    Setting up the initial control system will be done using an adjustable load in lieu of an inverter & battery bank. That will keep the cost down while resolving issues that come up. Getting the controller to start & stop PV strings in response to a changing load should be interesting.

    After I prove positive control, then I'll buy the Islanding inverter and a battery bank. I may try one of the 240VAC single phase inverters to see if the enphase units will sync up. I don't know if they need the neutral for syncing up. If that worked then a modest battery charger could handle recharging duty.
  • RCinFLA
    RCinFLA Solar Expert Posts: 1,484 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter

    Assuming you don't want 100 amp pushed at batteries, you are probably better off to regulate at a lower backfeed rate. Something like 3 kW or less. This gives you some headroom to account for AC loads suddenly being switch off that step functions the backfeed power. The headroom may give you more time to react.

    You are going to have to experiment to see how close to the max liimit edge you can run based on inverter overload time limit and how fast you can react to changes.

    It is not just how fast the processor runs, it is the sensors, ADC, interrupt priorities, etc. Trying to regulate near the 6kW backfeed limit leave little margin time to react.

    For AC coupled backfeed, in an offgrid condition, the XW is acting like an AC shunt regulator to maintain 240 vac line voltage.

    I assume you understand the XW's AC coupling software and how it slowly slews the AC frequency to shed the GT inverters when battery reaches charge limit voltages. During the backfeed push, generating higher DC charging current, if the batteries are too small their voltage will rise up even if not fully charged that may cross the XW's battery charge complete voltage setting so it starts a freq slew to shed GT's. This is another variable your software needs to take into account. You probably want to know when XW is about to do this. Monitor battery charge state or AC frequency may be the way to do this. In your design approach you probably want to avoid XW's GT shedding activity in most of the time.

    One of first things I would evaluate is how long it takes the micro-inverter to reconnect. Typical is 3 to 5 minutes. This is going to present you with some issues, which have no real solution other then giving up some PV yield in the least undesireable solution, and causing the XW to shutdown in most undesireable solution. It all depends on how agressive you try to be in keeping panels on line.
  • RK_Solar_Hopeful
    RK_Solar_Hopeful Solar Expert Posts: 69 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter
    RCinFLA wrote: »
    Assuming you don't want 100 amp pushed at batteries, you are probably better off to regulate at a lower backfeed rate. Something like 3 kW or less. This gives you some headroom to account for AC loads suddenly being switch off that step functions the backfeed power. The headroom may give you more time to react.

    You are going to have to experiment to see how close to the max limit edge you can run based on inverter overload time limit and how fast you can react to changes.

    It is not just how fast the processor runs, it is the sensors, ADC, interrupt priorities, etc. Trying to regulate near the 6kW backfeed limit leave little margin time to react.
    I had thought about this, but when you toss it out there all at once it does seem a bit questionable...
    For AC coupled backfeed, in an offgrid condition, the XW is acting like an AC shunt regulator to maintain 240 vac line voltage.
    I spent time looking for an AC shunt regulator until I understood what you meant. The XW will pass all the current it can to maintain voltage, until something gives. The control system has to be the regulator.
    I assume you understand the XW's AC coupling software and how it slowly slews the AC frequency to shed the GT inverters when battery reaches charge limit voltages. During the backfeed push, generating higher DC charging current, if the batteries are too small their voltage will rise up even if not fully charged that may cross the XW's battery charge complete voltage setting so it starts a freq slew to shed GT's. This is another variable your software needs to take into account. You probably want to know when XW is about to do this. Monitor battery charge state or AC frequency may be the way to do this. In your design approach you probably want to avoid XW's GT shedding activity in most of the time.
    I think I understand... but I will start small to limit the size of potential undesirable events.
    One of first things I would evaluate is how long it takes the micro-inverter to reconnect. Typical is 3 to 5 minutes. This is going to present you with some issues, which have no real solution other then giving up some PV yield in the least undesirable solution, and causing the XW to shutdown in most undesirable solution. It all depends on how aggressive you try to be in keeping panels on line.
    This area I gave a bit of thought already. That is why I'm starting with my basic control system concept, it flat out might not work. Then again, I might might get lucky and it will.
  • RCinFLA
    RCinFLA Solar Expert Posts: 1,484 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter

    With multiple inverters (four in your case) on a string, I wonder if there will be variability in time to connect between inverters causing the string power to come up in inverter power increments. Just another mushy variable.
  • RK_Solar_Hopeful
    RK_Solar_Hopeful Solar Expert Posts: 69 ✭✭✭✭
    Re: Grid Tied AC coupling with Islanding Inverter
    RCinFLA wrote: »
    With multiple inverters (four in your case) on a string, I wonder if there will be variability in time to connect between inverters causing the string power to come up in inverter power increments. Just another mushy variable.
    There most likely will be some difference in timing. I'm hoping they used higher precision components to minimize the differences. I should be able to manage around a couple of strings with wide timing. Another item to add to the testing list... sigh...