Adding storage to my Enphase system

[400bird]

Who has an inverter the utility can access and control (by means other than line voltage/frequency) at this time?

To go along with that, what would be the control method? Voltage and frequency are the only options currently available (as far as I know)
There sure is a lot of fud around this supposed big brother pushing charge commands. In my opinion, the most likely cause of GXMnow's issue is either a momentary grid disconnect or a PLC issue. He even admitted as much.

wheisenburg's issue is likely something different.

Can you both restart/power down the Insight? My Insight has had issues in the past where the grid alerts/faults didn't populate until after a power cycle.

 

[Hedges]

To go along with that, what would be the control method? Voltage and frequency are the only options currently available (as far as I know)
There sure is a lot of fud around this supposed big brother pushing charge commands.

California CEC list (if you download spreadsheet; not all columns on web page)

Solar Equipment List - CA Energy Commission

solarequipment.energy.ca.gov

Shows "Inverter CSIP Conformance" and the dates inverters complied with it.

"SunSpec Common Smart Inverter Profile (CSIP) Conformance Test Procedures"

Lists things like "DNS", IPv6, XML which would be internet communications with the inverters.

https://sunspec.org/wp-content/uploads/2018/05/SunSpecCSIPConformanceTestProceduresV1.0.pdf

I've read before that it had not been worked out who was responsible for cost of maintaining internet connection to inverters, but such connectivity was the plan.

 

[GXMnow]

Can you both restart/power down the Insight? My Insight has had issues in the past where the grid alerts/faults didn't populate until after a power cycle.

Wow, you called it.

I just power cycled the Gateway. It seemed to take a really long time to come back up, but it finally did. The Battery summary graph shows the blank spot where it didn't store any data. It was 4 minutes from power down to back online. I clicked on the "Events" tab, and it was still blank for current, but I clicked on history, and 2 new entries showed up. And yup, exactly at the 2 times my system started charging.

March 12th at 17 minutes after midnight, and again March 14th at 4:31 am.
Both are a grid voltage too high, error number 33.

So that solves this mystery, it is not my PLC programming or Big Brother. The grid voltage just crept up too high. At this moment, I am exporting about 40 watts, basically nothing. My grid voltage at the XW is a little high at 243.2 from L1 to L2. I have to wonder how high it went yesterday, when my Enphase panels were pushing 3,200 watts out to the grid?? That is when I would expect a high grid voltage error, not in the middle of the night when all the PV solar panels around me are doing nothing. It does seem odd.

@Hedges I always assumed that some type of hardware bridge would be needed to give them control, or I would have to supply the MAC address of my gateway etc. That is why I doubted it was an external control signal, but I just couldn't figure out what happened. Could the Gateway have powerline networking? How is the utility getting the data back from the smart power meter? I have read it uses a form of Zigbee, but that is a short range signal. Do they drive around and scan them from the street? Is there a gateway/hub/router on each block?

 

[400bird]

California CEC list (if you download spreadsheet; not all columns on web page)

Solar Equipment List - CA Energy Commission

solarequipment.energy.ca.gov

Shows "Inverter CSIP Conformance" and the dates inverters complied with it.

"SunSpec Common Smart Inverter Profile (CSIP) Conformance Test Procedures"

Lists things like "DNS", IPv6, XML which would be internet communications with the inverters.

https://sunspec.org/wp-content/uploads/2018/05/SunSpecCSIPConformanceTestProceduresV1.0.pdf

I've read before that it had not been worked out who was responsible for cost of maintaining internet connection to inverters, but such connectivity was the plan.

Good point. I guess I should have been more specific.
Unless you've made "them" aware of your inverter and it's grid connection, I'm not seeing a way for "them" to control the inverter.

 

[400bird]

Wow, you called it.

I just power cycled the Gateway. It seemed to take a really long time to come back up, but it finally did. The Battery summary graph shows the blank spot where it didn't store any data. It was 4 minutes from power down to back online. I clicked on the "Events" tab, and it was still blank for current, but I clicked on history, and 2 new entries showed up. And yup, exactly at the 2 times my system started charging.

March 12th at 17 minutes after midnight, and again March 14th at 4:31 am.
Both are a grid voltage too high, error number 33.

High voltage at 4 am? That's odd, I get higher voltage around solar noon, when the neighbors and I are all pushing to the grid and they're out of the house, not using as much energy. Meanwhile, I'm charging batteries...

So that solves this mystery, it is not my PLC programming or Big Brother. The grid voltage just crept up too high.

I guess you could bump the upper limit up just slightly.

At this moment, I am exporting about 40 watts, basically nothing. My grid voltage at the XW is a little high at 243.2 from L1 to L2. I have to wonder how high it went yesterday, when my Enphase panels were pushing 3,200 watts out to the grid?? That is when I would expect a high grid voltage error, not in the middle of the night when all the PV solar panels around me are doing nothing. It does seem odd.

@Hedges I always assumed that some type of hardware bridge would be needed to give them control, or I would have to supply the MAC address of my gateway etc. That is why I doubted it was an external control signal, but I just couldn't figure out what happened. Could the Gateway have powerline networking? How is the utility getting the data back from the smart power meter? I have read it uses a form of Zigbee, but that is a short range signal. Do they drive around and scan them from the street? Is there a gateway/hub/router on each block?

Good question, I assumed they installed hubs in neighborhoods. I'm interested to hear the answer to how the smart meter signal is read

I don't think it can be power line to the insight/gateway. Mine doesn't have an AC connection. I don't know think it can be power line to the XW, as it's not advertised. Also, I think the grid would get pretty messy if they were trying to put power line communication out to many houses across the grid.

 

[Ampster]

I'm not seeing a way for "them" to control the inverter.

Other than frequency or volts, that is the only way I know of. There is some kind of communication specification but I do not know how that is implemented.

 

[Ampster]

Good question, I assumed they installed hubs in neighborhoods

I thought each meter was a part of some kind of mesh nestwork. My gas meter and my water meter are wireless.

 

[Hedges]

I suppose bootleg equipment could be discovered, if on the network. Wouldn't know what customer, unless something about IP address of the router.
Sunny Web Box (which has a vulnerability, fixed admin password), was found in large quantity in a scan.

But permitted installs with utility PTO would now be of a model with SunSpec controls over Ethernet. However we aren't required to put it on the network, at least not now.

If someone had a backup system, utility might want to control inverters on the grid, but while yours is operating off-grid it shouldn't be controlled.

I'm hoping mine will be exempt, grandfathered as an installation before connection to internet required. But there is a lot of capacity out there, which could help move the grid.

Networking is also how zero export gets done by SMA. The Sunny Boy Storage I'm working with uses WattNode to sense power at utility connection, can shave peaks, and could command Sunny Boy to curtail output from PV. It doesn't look like SB by itself has a way to read meter or CT, that can be done by a separate controller. But SBS can do that, e.g. recharging off peak and seeking to zero import on peak.

 

[wheisenburg]

Well I rebooted my gateway. No events were on there. My AC voltage is running right around 250. Obviously with no export right now. That seems pretty high, but I am no seeing any AC over voltage events.

My DC voltage is at 53.01 so it is just about ready to enter a charging cycle.

 

[GXMnow]

I know my XW-Pro is visible on the internet as I can look into it, and Schneider has as well. I could take it off of the internet if I wanted to block it, but I see no reason at this point. It is nice to be able to check my battery status when I am away. They even warn about Modbus TCP not having any security. Once connected into my LAN, anyone knowing the IP address of my Gateway could control it.

My DC voltage is at 53.01 so it is just about ready to enter a charging cycle.

What is pulling your voltage down? How long has it been since the inverter stopped running at 53.5 volts?

My lithium ion NMC pack would take months to come down that 0.5 volts to trigger a recharge. And I would expect an LFP pack to take even longer with their super flat discharge curve. In fact, when my system is running low, each time it stops inverting at my 50.5 volt shut off point, the voltage rebounds and it ends up going back into Grid Support again about 4 times over the next hour before it will stay low enough. Then it will just hang at that voltage with no static load on it. I never had it actually "Recharge" just from the voltage falling. Early in march, we had a few cloudy days in a row and I let it manage itself to see what it would do. The morning of March 7th was interesting.


Shortly after 5 am, you can see the voltage dip and then the current goes from -24 amps up to zero as the inverter stops the first time. It did it 3 more times to about 5:35 am. Had it been totally dark out, that is about where it normally stays. But even with the cloudy sky, the Victron charge controller started to push a little current. That small charge current was enough to make the XW power back up like 18 times over the next 2 hours. The sun was also hitting the Enphase panels, so the load on the inverter was being reduced, but the AC solar was still not more than the house load, so it still shut off a couple more times until there was finally more solar coming in than grid power after 8 am.

I looked back a bit and I can't find another event where the voltage dropped to shut down with no solar coming in.

 

[GXMnow]

So I have a new issue...



I got up this morning and I like to check how low the battery went. WOW!!! What happened?

I doubt this was my PLC. It does not try to adjust charge current before 6 am. This started happening shortly after 5 am, I was still asleep. Discharge current shot from 9.44 amps to 74.33 amps. At 53.26 volts, that works out to over 3,900 watts at 5:07 am. The current tapers off to 51 amps after 3 minutes, but then the XW switches to charge mode, pushing 19 amps. But then it swings more and more for 90 minutes. The peak discharge rate hit 88.48 amps at 52.38 volts for over 4,600 watt at 6:13 am. It was just a short peak, but if only my normal overnight house load was running, that could be an export of close to 4,000 watts. That may have just exceeded my 16 amp export limit. It was short, and the 20 amp breaker obviously didn't trip. The highest charge current peak hit 56.64 amps 53.34 volts just over 3,000 watts at 6:15 am, that is after my PLC would normally start adjusting the current. The last of this event ended at 6:42 am and the PLC resumed proper export power control and zeroed the export power properly again. And it smoothly transitioned into charging current control as the sun rose about 8:40 with no sign of any issue.

And once again, there is no entry of any issue in the event log. But it sure looks like the grid was coming on and off that whole time. But with no solar coming in, it was going into charge each time the grid came back? But then it stayed out of charge once the grid was stable? That's not what it usually does? This is just odd?

I just looked at the "Energy Comparison" chart.


This only shows the full hour totals, but it is still pretty telling. The output back to the grid went from just 66 watt hours in the 4 am hour to 792 watt hours in the 5 am hour and 698 watt hours in the 6 am hour. But then back to nothing in the 7 am hour. Grid input (consumption) also went way up, from nothing overnight to 639 watt hours at 5 am and 691 watt hours at 6 am. But if you look at the net grid power, the total for the 2 hours is still just 160 watt hours, or 0.16 kilowatt hours, so the total power was really nothing.

The battery energy is really odd.
My system typically discharges between 700 and 1,200 watts through the night as things cycle on and off. So all the hours up to 4 am are in the range, discharge only, no charging. But then the 5 am hour has a discharge of 1.5 KWHs AND a charge of 0.448 KWHs. Net just over 1 KWH of discharge, which is about normal. And the same happened in the 6 am hour. 1.4 KWH of discharge, and 0.508 KWHs of charge.

I will do another Gateway reset and see if anything shows up in the event log.

 

[GXMnow]

Completed the Gateway reboot at 10:37 am.

No new event entries. Last events are the 12 and 14th that triggered the charging. So odd.

 

[400bird]

Someone else here reported a similar issue. Mine did it once in the wee hours of the morning. Something like 2 am and lasted about 2 hours, just like yours.

I have no idea where to go looking. I'm thinking it has something to do with grid frequency. Maybe the XW reacted to the grid frequency changing?

 

[GXMnow]

So this was interesting.

It was about 5:15 pm, it is still sunny, but at a very low angle, the Enphase panels are making just a tick more power than the house needs, so I am exporting a small amount of power, looks like about 500 watts right now. The XW is showing in "Grid Sell" but it's not selling anything. Here is the Status page in Advanced mode


You can see my current imbalance with L1 exporting 4.3 amps while I am importing 1.7 amps on L2. Sitting basically idle, the battery current is just 48 watts. The Victron charge controller is sitting in Float, pushing about 300 watts, so the batteries are seeing a small bit of charge current still.

While I was looking at this page, my eye caught a green flash in the grid of "Active Grid Support Features", but that first time it was too shar ofr me to figure out which one lit up. I watched it for a few minutes and saw it happen 3 more times. It lit up the top middle one "Freq-Watt Under Freq". With the page scrolled as I had it, I can see the grid frequency in 2 places, and both of them never changed from showing just "60" with no decimal showing. So I am not really sure what all is going on there.

Not sure if this will be the final layout, but I think it looks pretty good. I added a strip of Panduit wire duct under the charge controller. The pipe through the wall pokes through the back of the Panduit, so it is totally hidden now.

The black cable out the bottom is the Spectrum cable modem COAX. It was just tacked to the wall, it goes up on the left side of the XW and across the ceiling joists. The grey lead dropping out on the left side is the battey temp sensor from the XW. It's on the top battery module in my old battery bank.

The Victron now has the #6 awg wire in all openings. It's a tight fit.


My neighbor to the east has a tall palm tree, it's now casting a morning shadow onto my DC array.



My tree you can see off the corner of the garage is no issue. It's shadow is in my driveway. I am hoping it gets a little warmer and I can complete getting the wire up on the garage roof all into the conduit and tied up nicely under the panels. I have all of the materials. But it is windy an cold on the roof except when I need the DC panels making power. If I get everything lined up, I think I can move the wiring into conduit in about 1 hours of down time. Once it is over 65F out on my roof.

 

[GXMnow]

Sure enough, right as I sent the previous reply, I look over and the "Freq-Watt Under Freq" was lit, and I caught it with a screen capture.



And this time, it stayed for a bit and is showing 59.9 Hz.

 

[svetz]

Does California have enabled the "request for renewable energy" by drooping the frequency? I thought that was still in the discussion stages? If so, do you get special rates?

Or, is this just a situation where the grid is low on energy causing the dynamos to slow due to the extra drag?

I still haven't seen anyone post what they were really worried about (needing to cut out residential solar which they do by over-frequency).

 

[GXMnow]

Does California have enabled the "request for renewable energy" by drooping the frequency? I thought that was still in the discussion stages? If so, do you get special rates?

Or, is this just a situation where the grid is low on energy causing the dynamos to slow due to the extra drag?

I still haven't seen anyone post what they were really worried about (needing to cut out residential solar which they do by over-frequency).

I don't think the frequency dips are anything intentional. My guess is it is just load not being compensated fast enough.

My XW just did a smaller version of the odd charge/discharge swings. But this time it was much shorter in duration and the current did not swing as far. As the sun is not up yet, it was inverting from battery pulling about 10 amps, about 550 watts to run the basic house loads. But at 6:14 am it switched to charging at 17.6 amps, then discharging at 64.7 amps at 6:15 am, only 1 minute later?? At 6:17, two minutes later, it's back to charging at 13.9 amps. Another 2 minutes, and it's discharging at 62.9 amps. Then just 1 minute, it goes to charging at 6.9 amps. Only another minute and back to discharging at 46 amps, then 2 minutes until charging at 30.6 amps. Then the final spike of discharging at 66.8 amps. And just 1 minute later, it is back under my PLC control, pulling about 12 amps to run the house loads again at 6:25 am.

The total time from going off the rails to back in control was under 12 minutes. I wish I was watching it a little sooner to see this in action. I noticed it about 10 minutes after the spikes stopped. But looking at the PLC data, I really doubt it was doing this. While it can issue commands in well under a minute, the rules I set for starting and stopping a charge cycle do not allow it to swing like that. I require a fairly large swing in the available solar vs loads before it will reverse like that. This must be happening directly within the XW-Pro/Gateway. My guess again is that it is trying to stabilize wandering grid voltage/frequency issues. As big as the charge spike were, they only put 76 watt hours into the battery, but took out more like 200 watt hours. In the 12 minutes, it was exporting battery to grid for 5-7 minutes while charging from grid for about the same total amount of time. It smooths the graph to full minute steps, so it is likely about 6 minutes of each. But the discharge power was double the charge power. Here is a zoom in of the graph.

 

[GXMnow]

I have a theory.

The utility is adjusting the grid voltage.

I have only seen this happen a few times, always in the early morning, and only when exporting to the grid. The way the XW inverter works, it is acting as a voltage source, using the resistance to the grid to produce the export current I am asking for. It's not really a current source like a typical grid tied inverter.

Think of this like DC power. The grid is a giant battery, and my XW is a tiny battery. There is a small resistance between them, and to produce an export current, my XW battery is at a tiny bit higher voltage. Lets say it is 0.1 ohms, and I am exporting at 3 amps, so the voltage difference is 0.3 volts. But then the grid voltage changes, increasing by 1 volt. Now, instead of exporting by 3 amps, I am importing by 7 amps. The XW senses this difference, so it ramps the voltage to adjust, but it over shoots by a little, causing a bit too much export for a second, and as it comes close to the right setting, the voltage again moves a little further. The process repeats. Any time the grid voltage rises, my system momentarily goes to charging, taking some energy from the grid. If the grid voltage drops, it will increase the export current, until it can adjust to the right level again. I think this is why it makes that wave, and is always hunting back to the current my system was set for. It never stays at the large import or export power, it it just a peak.

What do you think? Does this seem likely?

P.S. It happened a second time at 6:30 this morning.

 

[Hedges]

Your inverter pushes power into the grid regardless of grid voltage, should only be maybe a glitch if grid voltage changes.
Except at some voltage limit where inverter is knocked offline, or does volts-watts.

UL-1741-SA allows wider voltage (and frequency) range, but has limited time ride-through.
Also allows optional ramp-down of power in response to frequency or voltage.

If any of that is happening, you ought to have inverter grid settings grid voltage parameters in the range where inverter is supposed to respond.

If your inverter is set for a particular amount of export, then the control loop could respond in mysterious ways.

I've watched my SBS ringing up and down with several second time constant.

 

[GXMnow]

I'll keep an eye on it, but so far, as odd as it looks, it is not causing a big problem. My battery bank has no issue dealing with the 80 amp charge swing. And as long as they net meter based on the full hour average, it is cancelling out.

I am 99.9% sure, it is nothing I am doing. It's not the PLC causing this. IT is something between the grid and the XW-Pro inverter.

 

[GXMnow]

Check this out.



It is shortly after 3 pm, the battery is full, and the sun is still shining strong. The Enphase panels are making a lot of power, so much that I am exporting over 3,000 watts to the grid. AC1 active power 3,016 watts, with 15.4 amps to L1 and 12.2 amps to L2. But, the frequency is low, so the XW is trying to "Help" it's pulling just 1.5 amps from the battery to export an additional 86 watts. When the "Freq-Watt Under Freq" turns off, the batery current falls back to zero, or even charging a tiny bit. But each time I see that flag light up, it pulls some battery current.

Here it is a couple minutes later, not correcting for frequency. A cloud also moved in so the back feed current dropped to just 1,342 watts.

DC power went from -86 watts to +44 watts.

If every XW on the grid is doing this, maybe it could help stabilize the grid.

 

[faccnator]

That has to be a setting right?

 

[GXMnow]

That has to be a setting right?

I am pretty sure that is part of the "CA Rule 21" Grid Code. The inverter was in "Grid Sell" mode, but I was only commanding it to export about 300 watts to cover loads in my main panel. Normally, about 20 watts extra would be going to the grid. But at that time, the AC coupled Enphase solar was already back feeding around 3,000 watts. Since my battery was already full, I dropped out of charge mode and with my grid tied inverters covering all mode loads and then a lot more, the XW-Pro battery inverter would normally just sit idle until the sun went down. But here it is, pushing out an extra 86 watts. A tiny drop in the bucket. But the AC frequency was low by less than 0.1 Hz. Any time the grid went back to 60.0 the current draw from the battery would stop.

 

[Hedges]

0.1 Hz shouldn't be curtailing based on Rule 21 or -SA.
Original UL-1741-SA considered grid in-spec from 59.3 to 60.5 Hz. I think the frequency-watts ramp down is from 60.5 to 61 Hz.

Variations in measurement error might have more to do with it.

In your screen shots, all I ever see is 60 Hz. But some have "freq-watt under freq" highlighted in green.

 

[GXMnow]

In your screen shots, all I ever see is 60 Hz. But some have "freq-watt under freq" highlighted in green.

In post #1,765 it dropped enough to show as 59.9 Hz.
I agree, it is going active very early. But I can see it happening. It is a very tiny amount of current. 86 watts at 240 volts is 0.358 amps. That is about my minimum static load back in my main panel. I just had a little fun since I wanted to heat up a Pretzel dog anyways.

Here is the screen, with my PLC monitor window. The PLC window is not very user friendly. In the simulated LCD window, you can see the Batt Volt at 57.670 and in the XW DC voltage readout, it rounds it to 57.7 volts. The Solar panels are nearly done making power, and I am starting to use some battery. This is my base load in my main panel, it is small. The XW window shows it as 88 Active Watts with a -0.9 power factor. My PLC asks just for Watts and it is showing 107. Battery watts is -670 in the XW window, and -664 in the PLC window. Again, the XW is rounding it off. I am not showing the lower section, Output of the XW. But the PLC is showing the output side running at 521 watts. That is my refrigerator, lights, this PC etc.

In the PLC green window on the left side, variable "U" is the current the PLC is telling the XW to send back to the main panel. In this snap shot it shows 391. That is milliamps of grid sell current.


And if you look in the XW window below th PLC screen, you see AC1 active current 0.4 amps, again, rounded from 0.391 to 0.4 amps.

Now I press start on the microwave oven, which is currently plugged into an outlet off of the main panel, not the backup loads. This places about a 1,700 watt load in the main panel.



Now we see the variable "U" in the PLC is now calling for 7,745 milliamps to the main panel. And if you look at the AC1 Active Current, there it is 7.7 amps. But that is at 240 volts. The microwave is only on L1. 10.5 amps are going out L1, and 5.4 amps are going out L2. Most of the L2 current is going to the utility transformer and coming back in to help power the oven at nearly 15 amps total on L1 in the main panel.

I am actually very surprised it works as well as it does like this. And the power meter totally ignores the phase imbalance. I have had it exporting over 15 amps on one leg, and importing over 14 amps on the other, and it shows just like 50 watts going out to the grid. I just hope they never change the meters to read the current of L1 and L2 independently.