diy solar

diy solar

~100kwh for 24/7 off grid home

tanoshimini

Solar Enthusiast
Joined
Mar 28, 2021
Messages
110
First, a picture:

21673BA2-8EF7-47F4-A3EE-B35497ED1359.jpeg

Now, some explanation:

We’re building a house up in Maine on a large, undeveloped lot (100+ acres, 50+ hectares). The site we chose to build on is well back from the road and our driveway is about 950’ (~300m) along it’s length. I was quoted an exorbitant price by the local utility to put in poles, string them, etc... all just for the privilege of paying a monthly bill and still suffering the occasional power cut. So, solar and batteries it will be.

I looked at Tesla’s Powerwall, but at north of $550/kwh, it’s just too spendy for the kind of sizes we’re looking for — and now, you can’t buy them without their solar tie-in. Shiny, pretty and new... but very, very spendy. Similar offerings from other companies had similar issues. We’re also going to need a system (or portion of it) up and in place on the property before we build the house in order to run tools and equipment. So, DIY it will be.

I’ve had a bit of experience (at a smaller scale) with this after putting 4 salvaged modules from a Model S into our sailboat some five years back (24v system), so it wasn’t a completely out-there idea for us to roll our own. That system is still functioning wonderfully, and I think it has a lot to do with the fact that we used the batteries lightly. While our small family lived on the boat for many years, the batteries almost never saw a DoD of more than 30% and we didn’t charge them to more than 4.1v. So, with this new setup, we’re going to want to try and oversize that, too, for all the same reasons.

Nowadays, the secondary market for lithium batteries is much more of a “thing” than it was when we did our boat up. More choices... so what to choose? I thought about buying an entire salvaged pack (or packs) from the Model S/X, but still couldn’t hit a price point I wanted... and rewiring their internals for different voltages is not something that I wanted to contemplate. The boat project was 24v, so the modules were “more or less” drop-in without much modification, but we had some problems with some DC equipment not liking the slightly lower voltages of the 6s arrangement of the modules (7s would have been better, but all of that was worked out eventually.) Doubling up modules to 12s leaves the voltage pretty light for a 48v system, with people seeming to favor 13s or 14s packs for that voltage range... and 48v seems like the way to go.

Batteries!

So, design goals:
  • 48v or higher, to keep the wire size down or to run cooler on the same mm² wire.
  • Oversized for the need, to keep the DoD shallow, cycling slower. Targeting 100kwh.
  • Cheap, so that I can afford to oversize the pack. Under $100/kwh would be ideal.
  • Outdoor, secure storage, because ? and ?.
  • 5 year service life, hopefully more.
  • Reusable parts, because change happens.
  • Modular, if possible, so that parts can be taken offline for maintenance or repair without disrupting life too much.
With that in mind, I looked all around and scoured the interwebs. After narrowing down the list, I decided to look deeply at the LG Chem UPB4860 rack batteries (48v 60Ah, derated to 40-45Ah) that have been hitting the market lately. I bought one from a local company and when I tore it apart, I was pleasantly surprised to see that it was structured in such a way (14s, divided into two 7s bricks, joined by a sturdy busbar w/ 300a ANL fuse) that they could be easily reorganized as as two giant 3.7v cells per rack unit. Rebuilt, charged and discharged several times in this new configuration, I found that they easily hit their 2.2kwh advertised capacity. Cool! At $220/ea, that’d be about $100/kwh. Now we’re talking. I bought 47 more, and managed to get them to drop the price by $40/ea. That’d put the total at 48 * 2.2 = ~105kwh, or just over $80/kwh. Outstanding!

Next, I’d need to work out where to put these things. My first thought was to store them the way they were built for, which was to use a server-rack of some kind. This would give me a sort of tall-ish profile, and I’d still need to insulate them from the cold... It didn’t seem like it was going to be workable without building a structure around them and that just ups the complexity of, well, everything and does nothing for ?. I tossed around a lot of ideas for various kinds of metal boxes and went back and forth for weeks, and then I happened to drive past a construction site and was struck by an epiphany: a jobsite box. I grabbed a tape measure and lit off to see what the internal dimensions of these things worked out to, since they come in a number of sizes from different sources.

After visiting several different stores and checking out a number of brands, I managed to find a 48x24x28” (~1220x610x710mm) one that would fit 8 units between the locking mechanisms with ⅛” (3mm) to spare, and room for 2” (~50mm) of foam insulation all the way around. I had been planning on using blocks of 7, but after reviewing the spec-sheets on the inverters on my boat (a pair of Victron Quattro 24/5000/120v), I noticed that I could go higher than 24v with them. I wondered if I could do the same with their 48v cousins, and sure enough, they’re rated to handle 66v. Eight of these rack units is 16s, so 60v nominal with a high voltage cut off of 4.1v/cell... that’s 65.6v. It’d all be within spec.

So, 6 jobsite toolboxes for 48 batteries, organized in blocks of 8, with 2 cells each. At $319/ea, the boxes aren’t too spendy... they’re completely reusable / resellable, heavy-gauge steel, designed for the outdoors, can be locked securely, are moveable by any machine with pallet forks and come in a lovely shade of blue. I had to hit four stores in the area to source them all. Watch out, though, not every one on offer was a gem... there were some I saw with dents or really ugly welds, etc. I left those and grabbed just the prettiest ones. We’ll put a concrete pad next to the house, and they’ll be two rows of three of them, looking all lovely and neatly spaced.

Running the numbers... with a large inverter at full continuous load, pulling 12kw from all of these batteries, that’d be 12000w / ~60v = 200a, split over 6 big batteries is ~33a to each, and each cell (of the 16 in each box) is made up of 7 trays of 4 pouches, so 33a / 28 pouches is ~1.2a per pouch, 4.8a per tray. The pouches were originally rated at 15Ah and 3C so even at full-load, that’s a sleepy 1.2a / 15Ah = ~0.08C. I’ll size the wiring as if the load will be two inverters at full tilt, even though the vast majority of the time we’ll be seeing a tiny fraction of that — wire losses should be very low. If my math is anywhere close to right, this will be a lovely, slow-paced retirement home for these old soldiers — may they live a long time.

In order to keep these things modular, I decided on using (genuine) Anderson SB175 connectors to transfer power. They’re rated at 280a wire-to-wire, and 200a wire-to-busbar and designed for 10,000 no-load connect/disconnect cycles. Conveniently, these job boxes have a well-placed 3” (75mm) cutout at the back for power cables, and the SB175’s are small enough to fit (2-1/4”, 55mm) through this opening. For now, I’m just going to pass them through, but these connectors are designed so that they can be hard-mounted. If I’m clever about it, I should be able to make the connector flush-mount with the outside of the box for a clean look, as well to avoid any protrusions that could get scraped off while moving the batteries around. Maybe 3d-print a nice weather-resistant cover plate.

More pictures!

0DDF6A86-CEF4-481A-BB2C-6552A6C09857.jpeg32637790-51F1-4ED0-B2F1-2AE4A8B81920.jpegFCC0D9C5-FAC5-428C-882B-C36C9C86640B.jpegDD07B3EE-F9E0-4F6F-B43B-942AB2E0CB6A.jpegF50DEF4E-1665-49E0-BFA0-0E60BC06FC31.jpeg67D5948E-9EE1-4574-9523-679D4E4AA680.jpeg4FC4F53C-84E5-4616-94C5-28B0EEE9D73F.jpeg6E13B4A8-A7AA-4D88-92A9-61109D2ACF2C.jpeg
 
Impressive. I really like the battery box design and it’ll be really slick with a flush mount plug. Strong work!
 
F84F36EA-1B8F-4295-90BB-7F9FDBBE2FF2.jpeg

Along with the move to 16s, I decided to buy the big-dog Quattro from Victron. 48/15000/240 (yep, 240v. more on that in a minute). Here’s my thinking (feel free to disagree!):
  • I want this part of the system to last a very, very long time — “grid-down” will be our every-day. Victron has been around for a while making good stuff for people that are more way finicky than I probably am, and they’ll probably still be around if I need them to service their warranty.
  • I can parallel the Victrons out to 6 units. I’ll probably never get past two, but if I need to...
  • They’re easy to work with. Programming them is pretty straightforward, even if the application you’re working on isn’t usual.
  • They’re made well. I know that these are actually made in a far-off land, probably in a factory across the street from a 1/2 priced competitor, but they’re made to specifications and they are tested to meet them.
  • They do what they say. If Victron is putting a number on their spec sheet, it’s far less likely to be marketing-smoke than an inverter I could source from aliexpress. I’m 100x more likely to get my money back from them if they are fibbing.
  • This is not the place to cheap out (for me). It’s going to be the the beating heart of my electrical system.
  • If I change my mind, the Quattro’s resale value will probably stay higher for longer.
  • I like blue. ??
Sticking with the job-box theme, I decided to buy the 60x48x28” (1525x610x710mm) version of the same blue job boxes to house the inverter(s). There’s room for two of these behemoths in there, with ample space for active cooling and/or insulation. By placing the inverter outside and nearer to the batteries, I can minimize the length/size of the wire that I’ll need to push all that current back and forth. By choosing the 240v, single-phase model I can minimize the wire size and voltage drop as I shoot power into our house, and eventual workshop. Since it’s the European version, they come configured for 50Hz, but 60Hz is just tweaking a setting. I’ll put an autotransformer right before my main service panel in the house and that’ll supply the split-phase / 120v. This will also keep all of the “weird stuff” outside of the house on a nearby concrete pad, which could (maybe) smooth out the eventual electrical inspection, since other than the autotransformer right before the main service panel, the house will be the same as a normal, grid-powered home in every respect.

51A15818-A869-4EFF-9611-70064D767784.jpeg790F7961-44F6-4595-897F-CAC0F5EED04B.jpeg
 
I, too, am looking forward to seeing this through with you. Thanks for sharing. My only worry so far is keeping a wary eye toward temperature management. It gets pretty cold in the north woods.
You make an excellent point! I haven’t gone into writing up the details (yet!), but I was planning on doing a full write-up on that in a post all by itself... but, the General Idea™ is to equip each of these boxes with four inexpensive ~25w silicone heating pads, wedged between the pairs of rack modules. I’d like to rig up a simple controller using an arduino nano and a relay, and have it talk to the BMS (over it’s serial port) to watch it’s low (and high) temperature alarms and react accordingly. I’m a big fan of open source, and I’m planning to release the schematics and the source for the controller on GitHub for other people to play with when I get it up and running. I love writing low-level protocol stuff, and it’ll be fun to do an implementation of the one used by the Daly BMS.
 
I don’t know if anyone is interested, but I’ve put together a google sheet where I’m keeping a running cost of all of the parts involved in the battery boxes, which you can find here. I’ll pretty it up — I promise. ? It lists costs for all of the major components, though I’ve left out little things like a few screws, bits of plywood and other little things that I had just laying around.
 
That’s a great question.
How did you come up with the 100kw target? Seems like a really big number but I guess it's not.

Here’s how I went about answering it for myself: I started out by looking at the average amount of power my current house uses in peak months and divided by thirty to give me a rough worst-case kwh/day figure. The average american household uses ~11kWh / day; we seem to use quite a bit more at about 30kWh / day. I pulled this data from the website for our current power company. Our new house will be a bit larger, but will be built out of much more efficient stuff/things... so I’m gonna guess that the difference will be a wash, and the final number will hopefully be a slight overestimate.

From the daily usage number, I tried to get a notion of what percentage of that usage wasn’t going to line up with solar very well, and ended up with a figure of ~10-15kwh per day. I could size the battery to that and charge/drain it daily, but I’ve learned from prior experiences, though, that one of the key factors for extending the life of lithium batteries is to avoid deeply discharging them — two other biggies being: lowering the voltage at which charging stops and avoiding extreme temperatures — more on all of that here. This led me to set a goal to use no more than 10-20% of the total capacity per day, which gives me the nice round number of ~100kWh.

(Over)sizing it this way does a number of things:
  • It ensures that even if we max out our (beefy) inverter(s), we’ll still keep the C rate low for each individual cell, which also extends the cell life by minimizing “side-reactions” in the battery that trap lithium and increase internal resistance.
  • We can maintain a healthy “buffer” against crappy weather — we’re not going to have an option for grid power and solar is not always guaranteed, and isn’t as plentiful in Maine as it is in other parts of the country.
  • If/when we use the generator, we can run it fully loaded (which is good for it’s longevity) without charging the batteries at high C rates.
  • If I need to take one of the (six) batteries offline for any reason (rolling maintenance, etc.), I don’t have to worry as much about disrupting everyday life.
The downside would be that it’ll take up more space and require more cells — but the downsides are only downsides if space is at a premium (we’ll have lots of outdoor space) or the cost of the cells is high (these cells were ~$80/kWh, and less than $120/kWh with all the fixin’s.)

Anyway, this is how I arrived at the number. I hope it helps in some way. :)
 
I don’t know if anyone is interested, but I’ve put together a google sheet where I’m keeping a running cost of all of the parts involved in the battery boxes, which you can find here. I’ll pretty it up — I promise. ? It lists costs for all of the major components, though I’ve left out little things like a few screws, bits of plywood and other little things that I had just laying around.
Thanks for the Google Sheet! I'm a fellow Mainer, let me know if you need a hand!
 
Anyway, this is how I arrived at the number. I hope it helps in some way. :)
Yes it did. I like to see the thought process behind these calculations as it gives me (and hopefully others) food for thought. I have 6kw and I am into it for about $150/kw, so it is good to see less expensive options.
The average American household uses ~11kWh / day; we seem to use quite a bit more at about 30kWh / day.
Last month was my lowest bill ever at 12.61kw per day. My house is a two story 2800sqft, but I will be downsizing to a single story 1500ish house soon. Hoping I can reduce that number by a lot. I really didn't pay much attention until late last year when I got into solar. I routinely pull 45-50kw a day in summer here in Texas with the heat, so it will be a challenge even in the smaller house, but I am hoping to mitigate that a bit with as much solar as I can squeeze into 2 acres. :p
 
According to the link you shared, your consumption is on par with the average American household.
10649 kWh/year ÷ 365 days/year = 29.18 kWh/day
Feeling pretty good vs Louisiana, not so good vs Hawaii :p

Louisiana had the highest annual electricity consumption at 14,787 kWh per residential customer, and Hawaii had the lowest at 6,296 kWh per residential customer.
 
Nice build. The Tesla Powerwall would not have been a good choice for off grid. It was primarily designed for grid interactive and would need a separate inverter to integrate solar and generator.
 
According to the link you shared, your consumption is on par with the average American household.
10649 kWh/year ÷ 365 days/year = 29.18 kWh/day
You're right. I made a mistake in not completely re-reading that article before I posted. I'm a dope. ;)
 
Nice build.
Thanks!

The Tesla Powerwall would not have been a good choice for off grid. It was primarily designed for grid interactive and would need a separate inverter to integrate solar and generator.
They’re known to be very price-competitive for pre-built systems, which is why I used them as a baseline for my cost model. I looked at the Enphase, LG, and a whole bunch of other systems as well. Tesla had them all beat on price per kWh. If the cheapest of the options was still too expensive for the size of the system that I wanted, I knew I was going to have to look into other options.
 
are you placing your batteries outdoors or in a detached shed?
TL;DR: These batteries will live out their days on a concrete pad about 50’ away from the house — they’ll be placed in a central location to shorten the wire runs between our house, workshop and well pump.

The boxes themselves are insulated on all sides with 2 layers of 1” of PolyShield EPS foam w/ a Mylar radiant barrier — any EPS or XPS foam will work, though. I haven’t decided how I want to insulate the lids yet, though more of the same would probably be a good start. I have also not yet added a heating system, though I will before the temperatures turn colder in the fall.

I try to be data-driven, so I’ve instrumented one of the boxes with a temperature logger. The unit can track two temperature probes, so I placed one probe inside the box, wedged into the center of the pack, and I taped the other probe outside and underneath the box to pickup ambient conditions. I left this in place with the boxes sitting in the sun/rain/etc. for two weeks in March to get a rough idea of how much the insulation and thermal mass of the batteries would smooth out the ambient day/night temperature swings. (I’ll post some graphs on that soon.) As expected, the insulation made a difference and kept the batteries ~10c cooler than the day, and ~7c warmer than the nights — much flatter temperature curves than for the ambient. We’ll see what happens as we get into the warmer / colder months!

I chose these particular boxes because they’re designed to live outdoors and the lid / lip of the containers are designed to keep water out. I added a ¾” strip of inexpensive compressible foam rubber weather stripping around the edge of the boxes to keep out any wind-blown water and bugs. I may add a bit of aluminum C-channel to the lid and mount a single 2’x4’ solar panel to shade the lid of the box as well as offset some of the parasitic losses of the heating / cooling / protection circuits. Other / different ideas are welcome, of course!
 
I have also not yet added a heating system, though I will before the temperatures turn colder in the fall.

I saw a video of some guy in Northern Vermont that uses a seedling mat to heat his bank. He has a wooden box enclosure with 2" of blue board. He has it rigged to turn on automatically, and it draws about 500wh per day. I'm curious what draw will be for a metal box and 1" of foam.

Also, do you mind me asking where you got these batteries from? I'd like to buy some from a local company as well, but I feel like I'm grasping at straws.
 
Back
Top