Wolf's PowIRwall

[Wolf]

Good work!

Did I just get an attaboy from @daromer ?
Thank you.
So after a reasonable solar day I was able to adjust the Batrium for final balancing. Man I do like the flexibility of the Batrium and the granular adjustments that can be made. I had 3 cells that had a tad too much voltage so manual burnoff needed to be made with my iCharger and a load resistor. Balancing is now complete and difference is at 0.02v, we will see how well it will keep.
Wolf

 

[OffGridInTheCity]

>Did I just get an attaboy from @daromer ?
You did! Very cool!

 

[Wolf]

Preliminary results are very good and promising After balancing all of the 42 packs to within 0.02v and running all night on them, not that my overnight draw is extensive yet, AC coming online soon, they remain remarkably well balanced.
I also updated Node-red to include the new packs and battery.
Wolf

 

[Wolf]

This my friends is a fully charged Battery Bank with 0.02v difference between all cells. I broke my arm patting myself on the back.
May 18th 2020 my frankenstein pack 14s80p went into production and so far has not had any degradation during its first year.
The addition of 2 more 14s80p batteries is now completed.
Time, dedication, research and the help from all of you has made this possible. The 4th and last battery is in the works and I will have a conservative 500Ah powerwall.
Once everything is done I will try to make a video tour of everything.
Wolf

 

[Wolf]

So 14 days into the addition of battery#3 and all is well. Pack voltage deviation is at a consistent level (Purple trace) between 0.018 to 0.030. Disregard the spike on 5/20 as I was updating node red to the latest and obviously there was no data so it reads 0V.



Also so far my max Amp discharge is 36 Amps 36A ÷ 3 = 12 A per battery 12A ÷ 80 = 150mA per cell. (My packs are 80p)

And my grid consumption has gone down drastically. Weather has been very good also but I am charging to 100% even on overcast days.

Wolf

 

[Solardad]

@Wolf Looks great! So what is now your overall kWH capacity vs. useable?

 

[Wolf]

Well if I go by the capacity tests of the packs which was done from 4.2v to 3v battery #1 was 186Ah, battery #2 207Ah, and battery #3 202Ah.
Considering the 21Ah difference between the low and high battery as relatively negligible and adding the Ah of all the batteries together coming up with a potential overall capacity of 595Ah. If we go by nominal voltage of 48v that would be 28.5 kWh. On the other hand if we use the max voltage of 58.8v we come up with 34.9 kWh.
Since my max voltage is set to 57.25 it ends up being 34 kWh if I were to run the batteries down to 42.0v which I do not. My cutoff is around 49v, 3.5v per cell so I calculated (educated guess) around 430Ah of usable which gives me a calculated 24.6kWh. Basically 70% of true potential.
That's what I come up with.

Wolf

 

[Solardad]

@Wolf so i am wondering with all this data that you are collecting have you thought about setting up an influxdb retention policy and/or a continuous query to downsample the timeseries data? I have just over a year's worth of data and plan on setting up some policies.

 

[Wolf]

@Wolf so i am wondering with all this data that you are collecting have you thought about setting up an influxdb retention policy and/or a continuous query to downsample the timeseries data? I have just over a year's worth of data and plan on setting up some policies.

@Solardad Not at this time anyway. I bought an 950GB SSD for this IBM Thinkpad for the purpose of storing the data for a long time. My first data was from IoTaWatt on March 28, 2019 and that database is only 2.0G. Total of all db's so far is 24GB bearly a dent on the 788G still available. At this rate I should be able to keep at least 20 years of data without changing the retention or downsample. I'll probably be in my 80s .
During this time I will more than likely move to newer and faster technology quantum memory or who knows what.
Wolf

 

[Solardad]

@Solardad Not at this time anyway. I bought an 950GB SSD for this IBM Thinkpad for the purpose of storing the data for a long time. My first data was from IoTaWatt on March 28, 2019 and that database is only 2.0G. Total of all db's so far is 24GB bearly a dent on the 788G still available. At this rate I should be able to keep at least 20 years of data without changing the retention or downsample. I'll probably be in my 80s .
During this time I will more than likely move to newer and faster technology quantum memory or who knows what.
Wolf
View attachment 25171

@Wolf Yeah, I did something similar to make sure I would not run out of space on my NAS (running a Docker instance as host for my stack) but the performance when pulling back historic data is horrible, I poll/collect data for all my data points every 5secs.. Roughly 2-3 days or less performance is decent, especially same day but larger timeframes is really not usable. I've decided that in reality 99% of the time I only look at sameday stats for the battery (batrium) and day over day or week over week for my house demand (IOTWATT). So I'll probably limit the batrium to 7 days of 5sec detail, rolling 30 of 5min blocks, 30-60 days 1hr blocks and then 60+ days purge. For house demand 14 days less 5sec then rolling 60 days 5 min blocks, 60-90 days 1hr blocks and 90-180 24hr block and then 180+ purge. I have some seasonal house demands that require (or at least my interest) to have the longer windows.

Actually now as I write up this recap now I wonder if influxdb supports indexes... that might help with my performance issues. hmmm. gotta check that out now.

 

[OffGridInTheCity]

@Wolf Yeah, I did something similar to make sure I would not run out of space on my NAS (running a Docker instance as host for my stack) but the performance when pulling back historic data is horrible, I poll/collect data for all my data points every 5secs.. Roughly 2-3 days or less performance is decent, especially same day but larger timeframes is really not usable. I've decided that in reality 99% of the time I only look at sameday stats for the battery (batrium) and day over day or week over week for my house demand (IOTWATT). So I'll probably limit the batrium to 7 days of 5sec detail, rolling 30 of 5min blocks, 30-60 days 1hr blocks and then 60+ days purge. For house demand 14 days less 5sec then rolling 60 days 5 min blocks, 60-90 days 1hr blocks and 90-180 24hr block and then 180+ purge. I have some seasonal house demands that require (or at least my interest) to have the longer windows.

Actually now as I write up this recap now I wonder if influxdb supports indexes... that might help with my performance issues. hmmm. gotta check that out now.

I collect/store (in MSSQL database) the charge controller data + Batrium Shanpshot Reports + temperature probes every 5 minutes. This is 288 rows of data / day and only 105,120 rows per year. 100,000 rows per year of indexed data is easily query'able.
Also, 5 minute intervals are perfectly adequate for visual reports such as this 'last 7hrs' of operation....


For 'live' - I use a custom web page + Batrium app - displayed on my 3rd montior 24/7

 

[Solardad]

I collect/store (in MSSQL database) the charge controller data + Batrium Shanpshot Reports + temperature probes every 5 minutes. This is 288 rows of data / day and only 105,120 rows per year. 100,000 rows per year of indexed data is easily query'able.
Also, 5 minute intervals are perfectly adequate for visual reports such as this 'last 7hrs' of operation....
View attachment 25192

For 'live' - I use a custom web page + Batrium app - displayed on my 3rd montior 24/7

@OffGridInTheCity i did not know about snapshot reports, very cool. I’ll have to check it out!

 

[FarmerJ]

@Wolf Might I be able to ask why you did cell level fuses on both positive and negative sides of the pack? Wouldn’t fusing just one be sufficient for the protection?

My original reason for asking this, cost, has proven quite ridiculous upon doing the math to realize that 80p would be a 5x16 making it just over 2ft of fuses at a cost of $3 a ft, but I am curious if there is some other reason to redundantly fuse both the positive and negative sides or if having both fused would increase the amps required to blow a single cell?

BTW my RC3563 finally arrived, and it is amazing how accurate a simple test and comparison to the IR cheat sheet can be at determining the cell quality. I wish I had known months ago to use a 4-wire IR tester. Now my YR1035+ Is going to my brother to use. Once I go through everything one last time I should be ready to build packs.

 

[Wolf]

@FarmerJ I used the cell level fuse sheet on both sides for ease of construction plain and simple. Once I came up with my bus bar design and the lexan side covers it was just easier to solder the bus bars to the whole sheet on both sides. Also I wanted to avoid soldering to the cell cans directly.
I would have had to at least on the neg side and string a wire from each cell to the bus bar. Total sturdiness of the pack was also improved. So all in all the expense for me was worth it. Total look of the pack had some influence also. Aesthetics you know.

The rated ≈ 5A fuse sheet does not increase the amperage at which the fuse would blow. Two 5A fuses in series still would blow at 5A one before the other im sure as the miniscule difference in the size and resistance would dictate which one would blow first.

I am glad and happy you found the IR cheat sheet and the RC3563 useful. Makes me happy to know my years of trying to persuade the group about the importance, ease of acquiring, and necessity of an IR check for a successful harvest. Nice to know when you stick a cell into a charger/analyzer you already have a good idea what the results will be.

Wolf

 

[Wolf]

Little update on the build. So we had a heatwave here in Maine, imagine that, 90°F for 3 days and very little relief at night. AC units ran pretty much non stop Solar was able to keep up and charge a little but in the long run the heat won to a certain extent. The Voltage threshold of 49V /3.5V per pack was reached and the inverters throttled slowly down to 0.0W. I was pulling power from the grid my heart felt heavy. It wasn't the money, as all in all I pulled ≈ 14kWh from the grid at a cost of $1.82, it was the idea that I had to succumb and hang my head in shame at knowing I had failed.

While battery #4 is still in the building stages I was wondering if with it I could have lasted another day for cooler weather and good sunshine to recharge the batteries. I also have another 750W of panels to put up but the heatwave came to early. Usually we don't have this heat this soon.

OK so what have I learned.
The Frankenstein battery did remarkably well with a V difference at this low of a voltage of ≈0.08. It certainly shows that mixing different cell from different manufacturers has an effect at the final cliff of the discharge curve. I will have to revisit the combination of cells in each pack and see if I can figure out a correlation. The 2 LGM26 (scooter/ ninebot harvested) batteries did very well with a very limited amount of pack voltage difference.
I also separated packs 1-14 and 15-42 out for agraphical 3 day representation Trace of 1-14 ▼

Trace of 15-42 much cleaner. ▼

I also did a bit of math and combined the output of my Arrays for the last 4 days and it produced 92.5kWh not too shabby. It did keep my house cool
Wolf

 

[OffGridInTheCity]

Yes sir - its so much fun to power your house. Was that 92.5kwh PV? or inverter output? How's your overall efficiency (PV in -> Consumption) looking?

The almost 0 difference for your other packs is a real testimate to what can be done. Nothing like operational confirmation!

My system (12.85kw PV - 45 panels) maxes out at 84kwh. In our very hot July/August months I need as much as 109kwh/day to maintain 72F/22C + the rest of house 'regular power consumption'... and I could do it with another 15 panels or so but the wife says "no way" - YOU PROMISED that 45 would be enough... (and its true I did promise)

 

[jdeadman]

Wow Wolf I do really like your data system. Here in Canaderp our heat waves are short lived and few and far between, and we just had 3 days of 29 and 30 degree C (86F) and we find that super hot. but my house only consumes about 40Kwh Max a day during the worst of it. and it's 1100sqft but it does cool at night to below 20C (68F) at night so I only Cool during the mid morning till after dinner.

 

[Wolf]

Was that 92.5kwh PV? or inverter output?

PV accumulation for 4 days. I have (18) 250W panels on roof 5 more going on shed soon. 3 to supplement house system and 2 to feed the sheds 12V electronics, fans, heat etc.

How's your overall efficiency (PV in -> Consumption) looking?

Good question I did run those calculations a while back and came up with some numbers. https://secondlifestorage.com/index.php?threads/wolfs-powirwall.7804/post-73962
I will have to revisit those and see if they still stand

I do really like your data system

Yea @jdeadman it's all done with smoke and mirrors ESP32s and lots of experimental sketches and code. A little luck every now and then.
Just added 2 more GTIL so excuse the mess.


Wolf

 

[jdeadman]

That looks busy. I'm currently only running a MPPsolar 2424lv unit for charging and inverting. It's currently running a fridge at camp offgrid with 1200W of solar. Need to up my data acquisition game

 

[Wolf]

That looks busy

Yes it does look busy. Lots of stuff happening there. Sensor collection, breakers with shunt trips on all the critical locations so if there is a panic all systems can be shutdown with the push of a button. and also overkill on all the cabling. Most everything is rated at least next AWG gauge up from minimum. If you need help sorting out how to acquire data let me know I got all kinds of sketches and ways to do it.
Wolf