Earth Notes: Second Enphase AC Battery Home Storage (2021-09-15)
Updated 2022-09-16.Solo
No van required; the EcoPartners installer turned up about in an estate car. At least two other jobs followed mine.
At about he flipped the breaker in the consumer unit for the Envoy (and battery). The final poll over the LAN before this was at :43.
(While he was working I noticed a bug in the script that captures readings every few minutes from the Envoy for logging and analysis. When the overall system is exporting to the grid a flag file is created by this script in the filesystem for other subsystems to observe. When not exporting, or when the Envoy can't be reached, it is removed. In fact, it was not being removed, and that is now fixed.)
At the Envoy was powered up again, and commissioning of the new battery started. The Envoy responded to a poll over the LAN at :23.
By it was all done, with both batteries showing in the Enlighten display.
Thankfully the firmware did not need to be updated, and so I had not lost access to the production.json API, ie I can still poll for data over the LAN.
Here follows a chunk from the Enphase log with the last old-world single-battery reading, then the new ones after a 40-minute gap. Note that storage.percentFull jumps from 22% to 25% to 30% suggesting the the new battery was shipped 1/4 to 1/3 full. Also note that the new battery takes a while to start charging, ~20 minutes in, then an aggregate 554W is hit (dropping to 539W) from 269W before.
20210915T09:43Z consumption.readingTime 1631698971 consumption.net.wNow -177.223 consumption.total.wNow 331.911 production.wNow 509.135 storage.percentFull 22 storage.wNow -269 storage.readingTime 1631698698 20210915T10:23Z consumption.readingTime 1631701381 consumption.net.wNow -359.412 consumption.total.wNow 299.013 production.wNow 658.425 storage.percentFull 25 storage.wNow -263 storage.readingTime 1631701277 20210915T10:28Z consumption.readingTime 1631701682 consumption.net.wNow -262.414 consumption.total.wNow 300.268 production.wNow 562.682 storage.percentFull 30 storage.wNow -263 storage.readingTime 1631701680 20210915T10:33Z consumption.readingTime 1631701980 consumption.net.wNow -354.156 consumption.total.wNow 286.988 production.wNow 641.144 storage.percentFull 30 storage.wNow -276 storage.readingTime 1631701825 20210915T10:38Z consumption.readingTime 1631702281 consumption.net.wNow -156.085 consumption.total.wNow 571.11 production.wNow 727.195 storage.percentFull 31 storage.wNow -274 storage.readingTime 1631702128 20210915T10:43Z consumption.readingTime 1631702580 consumption.net.wNow -160.17 consumption.total.wNow 639.929 production.wNow 800.099 storage.percentFull 32 storage.wNow -508 storage.readingTime 1631702427 20210915T10:48Z consumption.readingTime 1631702880 consumption.net.wNow -477.728 consumption.total.wNow 633.996 production.wNow 1111.725 storage.percentFull 34 storage.wNow -554 storage.readingTime 1631702791 20210915T10:53Z consumption.readingTime 1631703181 consumption.net.wNow -310.84 consumption.total.wNow 649.714 production.wNow 960.554 storage.percentFull 36 storage.wNow -539 storage.readingTime 1631703093
Installed
The new AC Battery was installed above the existing one to avoid taking up any more useful space. Yes, I did cover the overly-bright indicator lamp! 
Screenshots
All Systems Go
By about the sun had helpfully appeared to show a sustained aggregate charge rate of ~540W across the pair.
At the batteries were reported to be 100% full at 2480Wh stored.
At the batteries where discharging at over 300W to avoid imports during cooking. Twenty minutes later demand (~3kWh) was high enough that imports were required but more import was now being averted than with one battery. Hurrah!
Note that the battery had emptied even before everyone went to bed last night. So we were importing all night and this morning even though we yesterday generated far more than we consumed. The extra capacity should reduce the incidence of this, ie improve self-consumption.
The Enlighten "energy independence" for September up until the evening of (ie just after) the installation is 70%. 2020-09 was 68%, 2019-09 was 73%, and 2018-09 was 74%.
2021-10-26: the additional oomph (technically, power) of the new pack means that I can bring water to the boil (slowly, on 6, on the small induction spot front right) without need for PV generation or grid imports. Solar noodles after dusk...
Off-Grid Too
As of I replaced my off-grid battery bank to give me a full 1kWh+ of usable capacity again.
It has already dealt with very bright days, very dull days, and dull days with random bursts of direct sunshine, steadily keeping about 0.25kWh/d of 'dump' load (the Internet router) away from the grid.
So I now have a little over 3kWh of automated electrical storage, one third of it off-grid. In total that is maybe half a day's use.
My target, covid-permitting, is to plan in 7kWh of heat storage for early spring next year. That is nearly two typical days' hot water.
Stats to Look Out For
See stats from batteries 3 and 4: November (intermonth) profile comparison.
Profile before and after: September
Below is a simple before and after grid load profile (weekdays) as I did last December, but before and after the Enphase upgrade on 2021-09-15. Days 1--14 for 'before' and 16--29 and for 'after', eg 2 full weeks each side. (Excluding the installation day itself.) 
The source data is from September's Enphase monthly net energy report.
The days have been getting shorter rapidly through this, and that would diminish any relative improvement in self-consumption. So any improvement remaining is more robust than otherwise. Also the data sample set size is small. (And the weather does its own thing, and so may confound all comparisons!)
Note that BST is currently applied to the Enphase data, ie running from midnight local time rather than UTC. I'll fix that for some other data analysis run!
Date/Time,Energy Produced (Wh),Energy Consumed (Wh),Exported to Grid (Wh),Imported from Grid (Wh),Stored in AC Batteries (Wh),Discharged from AC Batteries (Wh) 2021-09-01 00:00:00 +0100,0,8,0,0,0,8 2021-09-01 00:15:00 +0100,0,17,0,1,0,16 ...
Results have been captured of a run of the script (with a copy of the script as run). This includes weekday, weekend and all filters.
It's difficult to see vast changes by eye with the graphs. On the weekday charts I'd say that if there is any difference then it's more prominently from increased capacity (more Wh) allowing energy carried overnight to better support morning (breakfast) loads. The green (net) line is below the purple (gross) line in the 'after' graph.
Profile last year vs this: October
Copying over the data and running the script for 2020-10 and 2021-10. In 2020 there was only one AC Battery, for 2021-10 there were two.
% vi img/PV/load-profile/202110/load_profile.sh % mkdir ~/tmp/load-profile % cp data/16WWHiRes/Enphase/adhoc/net_energy_202{0,1}10.csv.gz ~/tmp/ % sh img/PV/load-profile/202110/load_profile.sh
This extra battery power and capacity seems to have had two effects:
- Shaving ~400W off the peak rather than ~200W last year, though the import peak is still just over 800W!
- Eliminating most other grid imports from 6pm to ~6am.
The source data is from October 2020's and October 2021's Enphase monthly net energy reports.
Results have been captured of a run of the script (with a copy of the script as run). This includes weekday, weekend and all filters.
Some other possible stats to monitor/analyse
- Change in self-consumption by month/year by my definition and/or Enphase since before any PV was installed.
- Days started or ended (eg at 00:01, 23:59) with no stored energy by month/year.
- Days (or hours) where battery reaches 0% (or 100%) by month/year.
2021-10-18T09:50Z: appended storage.whNow to Enphase 5m stats log to make future analysis easier; should really have done this just before this second battery went in.