LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[toddedger]

That's awfully basic. I'm sure it's covered in the electrical facts post. Basically the fuse size is determined by the size of the wire following it. If there is a short in the wire, you want the fuse to blow before the wire heats up and catches on fire. Wire size is determined by the load you will put on it. You have to take into count the wires load capacity, and the voltage drop down its length. There are tables usually published by the wire manufacturer that show this information.

[CarstenWL]

... I am fully aware of the basics, I even know the norm DIN EN ISO 10133 ;-) ...

I was amazed to see some suggestions like the one attached with 2x 400A fuses, thats why I started to 'think about if I am missing something'.

Thanks,

Carsten

[transmitterdan]

There are 3 critical fuse specs:

Max DC voltage
Max interrupting current
Fuse opening current

The max interrupting current ideally would be 10,000A IMO. Some will say 5,000A is ok but for Li banks I believe it is marginal.

[OceanSeaSpray]

Originally Posted by transmitterdan

There are 3 critical fuse specs:

Max DC voltage
Max interrupting current
Fuse opening current

The max interrupting current ideally would be 10,000A IMO. Some will say 5,000A is ok but for Li banks I believe it is marginal.

You need to remember that in order to produce such high short-circuit currents, you also need to have an unbelievably low short-circuit resistance with a circuit capable of holding - where are you going to find that on board??
A low-voltage accidental short doesn't exactly look like a solid copper bar bolted tight across the terminals.

There is a lot of hype about fusing requirements, but practically just fuse at twice the maximum current you ever expect to see on the installation. In most cases, it should be anything between 250A and 500A fuse rating.
Most shorts start relatively small, with high resistance, create heat and further damage insulation, become worse as a result and finally blow the protection. The higher you fuse, the worse off you are, and don't imagine for a second that a sustained short at 50A can't start a fire.

Next, if you don't size up your feeder cables to carry 10kA at low voltage drop, it also helps enormously. They need to be big enough to give low voltage drop at normal full load and carry the fault current until the fuse blows - no more.
This is why the earth core in a big feeder cable is a lot smaller than the phases.

Practically, the size required for a cable increases with its length. There is no practical direct relation between cable size and current alone and bigger is better only until you start thinking about the potential for fault currents.

[transmitterdan]

Originally Posted by OceanSeaSpray

You need to remember that in order to produce such high short-circuit currents, you also need to have an unbelievably low short-circuit resistance with a circuit capable of holding - where are you going to find that on board??
A low-voltage accidental short doesn't exactly look like a solid copper bar bolted tight across the terminals.

A 1.5 milliohm short is not impossible to get and will produce more than 5kA from a healthy Li pack. There are many fuses commonly available that may not open under those conditions. I think all big Li packs should have the main fuse with 10kA interrupting capability. The actual fuse opening rating should be based on the expected max load and wire size.

[OceanSeaSpray]

Originally Posted by Maine Sail

I have only noted this once and that was when I went 80 cycles between a full discharge capacity test. I kind of panicked when the capacity suddenly dropped off by about 20 Ah's from only 80 cycles before. After the discharge and full recharge all the capacity was still there and nothing to worry about.. I suspect because I have been capacity testing every 50 cycles that I am not seeing the false ceiling effect quite as much.

Interestingly enough I have scoured the literature and not found any white papers dealing with this...? If anyone has one please let me know.. Still I continue to charge to 14.0V or less and the bank is still producing in excess of its rated capacity when discharged at a load well in excess of our average house loads.

I would have fully expected these cells to drop off a bit by now, they are 2009 cells, but they keep on chugging...

I have read at least one paper about memory effect in LFP cells and it showed that the charge voltage curve after a fractional cycle had a slightly different shape (see here). It didn't result in reduced capacity, because they just pushed through it by going to the normal end-of-charge voltage, but I think that, in the case of low-voltage charging, it has the potential for producing exactly what Terry is describing: you hit the bump in the charge curve and it causes the charge to end prematurely. Then you make it worse by cutting off earlier and earlier as the bump moves back against you. It is a pretty remarkable observation.

Note that charging to 14V doesn't in itself mean a partial charge. It only does if you don't absorb. Lower voltage just means lower stress and slightly longer charging time, it doesn't mean you can't fully charge or overcharge.

After a number of experiments, I now implement an absorption time at a modest voltage (usually 13.8-14.1V) and I use the higher values for high C-rate sources because of the internal resistance effect.

[CarstenWL]

... thanks transmitterdan and OceanSeaSpray!

Carsten

[OceanSeaSpray]

Originally Posted by transmitterdan

A 1.5 milliohm short is not impossible to get and will produce more than 5kA from a healthy Li pack. There are many fuses commonly available that may not open under those conditions. I think all big Li packs should have the main fuse with 10kA interrupting capability. The actual fuse opening rating should be based on the expected max load and wire size.

Good luck with getting 1.5mOhm. The total measured internal resistance of a new 12V nominal, 400Ah pack I was working with recently was already 1.67mOhm. As I said, by the time you go through the various bolted connections, cable lugs, shunt, disconnect switch, some length of cable etc, you do have a lot more resistance than you would think and it is the total circuit resistance - battery included - that matters.

If you fuse at the expected maximum load (which is never done for good reasons), you are going to blow that fuse one day due to surges and on top of this, you will also find that you are dropping too much voltage over it at full load. Fuses get warm long before they blow.

[OceanSeaSpray]

The original article about memory effect in LFP cells was published in Nature Materials in 2013.

Eric

[Maine Sail]

Originally Posted by OceanSeaSpray

I have read at least one paper about memory effect in LFP cells and it showed that the charge voltage curve after a fractional cycle had a slightly different shape (see here). It didn't result in reduced capacity, because they just pushed through it by going to the normal end-of-charge voltage, but I think that, in the case of low-voltage charging, it has the potential for producing exactly what Terry is describing: you hit the bump in the charge curve and it causes the charge to end prematurely. Then you make it worse by cutting off earlier and earlier as the bump moves back against you. It is a pretty remarkable observation.

Note that charging to 14V doesn't in itself mean a partial charge. It only does if you don't absorb. Lower voltage just means lower stress and slightly longer charging time, it doesn't mean you can't fully charge or overcharge.

After a number of experiments, I now implement an absorption time at a modest voltage (usually 13.8-14.1V) and I use the higher values for high C-rate sources because of the internal resistance effect.

I am absorbing a 400Ah pack down to about 5-10A at 13.8V. The pack now has over 700 cycles most to 80% DOD. I only noted this memory effect once and have not been able to reproduce it since but it certainly seems plausible if you just stop at 13.8V. I will dig into that paper later... I have bounced around between 14.2V and 13.8V and have settled on 13.9V & 13.8V which yields a minimal absorption time at our charge rate. Our solar is set to 13.7V but we rarely need it because the bank charges so fast that even short runs in and out of a harbor with the engine running yield more than a days supply of energy.

[hellosailor]

dan-
" The actual fuse opening rating should be based on the expected max load and wire size."
I would suggest rephrasing that, i.e. " The actual fuse opening rating should be based on the expected max load and wire size, whichever calls for the lower fuse value in order to be protected." Depends on what you're trying to protect, and whether there's just one load involved.


But last time I checked, if a fuse wouldn't blow under a 5kA load, there was no way in hell that putting in a 10kA fuse would make it blow any better of faster. The small fuse should blow faster, and potentially even explosively, well before the larger fuse opens. And once the fuse has blown--unlike a breaker, there's nothing left to weld itself shut. Or is there?

[OceanSeaSpray]

Originally Posted by Maine Sail

I am absorbing a 400Ah pack down to about 5-10A at 13.8V. The pack now has over 700 cycles most to 80% DOD. I only noted this memory effect once and have not been able to reproduce it since but it certainly seems plausible if you just stop at 13.8V. I will dig into that paper later... I have bounced around between 14.2V and 13.8V and have settled on 13.9V & 13.8V which yields a minimal absorption time at our charge rate. Our solar is set to 13.7V but we rarely need it because the bank charges so fast that even short runs in and out of a harbor with the engine running yield more than a days supply of energy.

That is pretty well absorbed already I would say, so it should explain the absence of memory effect. I have stopped a 400Ah bank at about 15A recently, it would have been down to your levels in a matter of minutes after that. My absorption times are around 30-35 minutes normally and they are important.

I also tend to cycle slowly and deeply. I feel it minimises the amount of time spent at high SOC. DoD is really the last thing I worry about. I have spent the whole winter between 20% SOC and 50% SOC without a glitch. All my power is normally solar and most of the time running the engine for any reason gets me near-fully charged in very short order.
If any memory effect developed over the winter, the bump wasn't high enough not to climb over it when I started getting good charging again and I didn't notice anything.

I am still thinking about solar strategies with the winter/summer conditions. In winter I basically keep charging all day, every day, and in average it offsets my consumption.
Another boat did the same, but using more power and ran the engine briefly about once every 2 weeks, so very slow cycling mostly at low SOC as well.

[Maine Sail]

Originally Posted by OceanSeaSpray

That is pretty well absorbed already I would say, so it should explain the absence of memory effect. I have stopped a 400Ah bank at about 15A recently, it would have been down to your levels in a matter of minutes after that. My absorption times are around 30-35 minutes normally and they are important.

This is just about where my 400Ah pack runs. I am sensing charge voltage right at the battery terminals.

Originally Posted by OceanSeaSpray

I also tend to cycle slowly and deeply. I feel it minimises the amount of time spent at high SOC. DoD is really the last thing I worry about. I have spent the whole winter between 20% SOC and 50% SOC without a glitch. All my power is normally solar and most of the time running the engine for any reason gets me near-fully charged in very short order.

We discharge to 70-80% DOD and generally only charge when we have an opportunity, and where ever it gets to SOC wise during that opportunity charging is fine by me.

Unfortunately I do cycle to full approx every 10th cycle because the Ah counter will not remain accurate beyond that. I have hundreds of hours of testing numerous Ah counters with LFP and they all FAIL to stay accurate over more than about 7-10 cycles. If I can reset it to full every 10 cycles or so I am in much better shape..

Originally Posted by OceanSeaSpray

If any memory effect developed over the winter, the bump wasn't high enough not to climb over it when I started getting good charging again and I didn't notice anything.

When my bank is not in use or not being aggressively cycled to match our boats behavior the bank is stored at 50% SOC at about 45-50F..

Originally Posted by OceanSeaSpray

I am still thinking about solar strategies with the winter/summer conditions. In winter I basically keep charging all day, every day, and in average it offsets my consumption.
Another boat did the same, but using more power and ran the engine briefly about once every 2 weeks, so very slow cycling mostly at low SOC as well.

I find the biggest hurdle is forgetting & unlearning everything we learned with lead acid. Once you get comfortable with never needing to get back to 100% SOC, nor really wanting to, nor caring if you hoover between 20% SOC and 60% SOC for weeks on end, managing LFP is significantly easier than dealing with lead acid.

For us it is extremely simple, when the bank is full we turn off all charge sources. When the bank gets down to 30% SOC or lower we will take an opportunity to charge when it presents itself. Only on rare occasions have we needed to turn the charge sources back on at 50% SOC or so in order to go 4+ days on the hook with peace & quiet. We can always extend our time without noise by flipping solar back on but I have really been trying to cycle to 80% DOD, or as close to it, with each cycle to see if all the promises of LFP are actually true, they certainly seem to be. If I had done what I have done with LFP over the last 4 years, with lead acid, I would have been through 4 banks already.....

[transmitterdan]

Originally Posted by hellosailor

But last time I checked, if a fuse wouldn't blow under a 5kA load, there was no way in hell that putting in a 10kA fuse would make it blow any better of faster. The small fuse should blow faster, and potentially even explosively, well before the larger fuse opens. And once the fuse has blown--unlike a breaker, there's nothing left to weld itself shut. Or is there?

Maybe you are confusing max interrupting rating with the rated opening current. Say you have a 250A fuse. If you run 300-400A then it should blow within a certain time. But some fuses if you run many times over the rating (say 7,000A) will never open because they cannot handle the massive current flow and they continually arc over and never perform their function. This is called the "max interrupting rating". Typical values for DC fuses are 5kA and 10kA.

[OceanSeaSpray]

Originally Posted by Maine Sail

...
I find the biggest hurdle is forgetting & unlearning everything we learned with lead acid. Once you get comfortable with never needing to get back to 100% SOC, nor really wanting to, nor caring if you hoover between 20% SOC and 60% SOC for weeks on end, managing LFP is significantly easier than dealing with lead acid.

For us it is extremely simple, when the bank is full we turn off all charge sources. When the bank gets down to 30% SOC or lower we will take an opportunity to charge when it presents itself. Only on rare occasions have we needed to turn the charge sources back on at 50% SOC or so in order to go 4+ days on the hook with peace & quiet. We can always extend our time without noise by flipping solar back on but I have really been trying to cycle to 80% DOD, or as close to it, with each cycle to see if all the promises of LFP are actually true, they certainly seem to be. If I had done what I have done with LFP over the last 4 years, with lead acid, I would have been through 4 banks already.....

Whenever I think about my bank, it is in terms of not charging it too high or too often... quite a paradigm shift indeed!
Most of the development work I have under way at present revolves around SOC management.

I have found I don't need any Ah counter, voltage is more trustworthy in the ranges where I am interested in knowing something about the SOC. However, I have managed to configure a few Coulomb counters to auto-reset in such a way that they keep reading something meaningful (but most likely not accurate) without intervention and the owners are happy.
This being said, I do have something in the pipeline to get a robust, maintenance-free SOC hopefully because I want to use it for bank management.

There is one situation where I am not very happy with not charging until the bank gets low, it is when power availability is scarce: then I want to always make the best of what I can get and extend the cycle as you mention...
This is why I want good real-time SOC data so I can get the system to make smart decisions and adapt to seasonal variations and consumption.
At the moment, I limit the SOC solar power is capable of achieving, but it is only a temporary measure until I can get things to work really properly.