There is so much to say on this topic that I struggle with where to begin. It's been an exercise in drinking from a fire hose, and everywhere I turn there is another hose blasting away. So I'll just pick a starting point, and describe what our plans are for 6837, and how I'm proceeding from here to there. That should set out the context, and I can write other articles later digging into particular aspects of Lithium power systems. Make sure you have read this Lithium Battery Backgrounder before reading this article, or some of the terms and concepts may not make sense. I'm hardly an expert, but learning fast, and my goal will be to distill it down to something more digestible, without at the same time dumbing it down. So here goes...
First a little personal background. I'm a retired electrical engineer/computer science person, turned exec, but still a geek at heart. Our boating is long distance/remote cruising with extended periods of time away from civilization, so dependable systems are vital, and repair/workaround essential when the inevitable issues arise. We are building another Nordhavn and the plan of record is to use LFP, but before pulling the trigger on LFP for the boat, I'm building a proof project in a slightly less demanding environment to test and assess LFP viability for the boat.
Now I already need to pause and explain something. You are probably wondering what "LFP" is. It's short hand for a particular type of lithium battery, specifically LiFePO4. There are many different types of lithium batteries, and LFP is the only on that I (and most other people) consider suitable for use on a boat, mostly for safety reasons. Some other time I'll dig into the different types and their characteristics, but not now.
I mentioned a test project, and that will be to build an LFP system for an off-grid house. In nearly every way the requirements are the same as the boat, but less consequential if there is a failure, and a lot easier to get parts etc when needed. If nothing else, the shipping address isn't changing all the time :-)
With that preamble, here's my current thinking for the LFP sub-system:
- I am building the system myself. It's not that I object to the available pre-engineered BMS (Battery Management Systems) systems, but I'm very concerned about the long term viability of nearly all of the companies and products. When some part fails in 3 or 5 or 10 years, I think there is a good chance that any given company I might choose will be gone, and I'll have to buy a whole new BSM. Victron is about the only company that I think has long term staying power, but the cost puts LFP into a price range where it isn't worth it to me. I want something that is long term repairable and supportable and will live the life of the batteries. So the plan is to an industrial Programmable Logic Controller (PLC) with off-the-shelf modules for monitoring and control. Pretty much everything in the modern world is run by PLCs. Manufacturing, lighting controls, building environmental controls, elevators, and many of the control systems on larger ships. They are made by many very large companies, have highly standardized interfaces and protocols, are incredibly robust and reliable, and are relatively inexpensive. And best of all, a lot of much bigger, much more important customers will be complaining way ahead of me if there are problems.
- The PLC "BMS" will be a monitoring and fail safe system, but not a balancing system. Instead, I'll follow the approach of operating the batteries away from their SOC end points, or "knees" in the voltage curves. When monitoring suggests an imbalance, I will have to respond manually to re-balance.
- For batteries I am using CALB CA180s. CALB seems to be the most reliable vendor, but it's admittedly a low bar, and I am open to alternatives. The 180 cells seem to be the largest that are readily available. Bigger cells means less paralleling, few interconnections, etc. The 400s are appealing, but don't seem to be available
- I'm planning on 360AH of LFP capacity operating at 48V, so about half of the FLA usable capacity that I currently have. That will last me 1.5 to 2 days without sun. If the batteries get too low, the generator will start and recharge in a little over 2 hrs. The generator may run more frequently, but for a lot less time in each run, and less in total. Right now, a full recharge of the FLA batteries takes just under 6 hrs.
- The battery configuration will be paralleled pairs of cells, wired in series for 320ah @ 48V. Short hand for this configuration is 2P16S. That's "blocks" of 2 batteries in parallel, designated by "2P". Then 16 of those "2P" blocks are wired together in series, designated by "16S". Put it together and you have 2P16S indicating 2 parallel battery groups, wired 16 in series, for a total of 32 batteries.
The other battery wiring option is two 48V strings of single 180Ah batteries (16S), with isolation switches for each string. So two strings - effectively 16S2P - totaling 360Ah at 48V. This later arrangement would be to provide for string isolation for diagnostics, and to be able to run on one string if the other is out of service. I might convert to this arrangement later on, and it might be the preferred arrangement for the boat for increased fault resilience.
- For monitoring, I plan to instrument voltage and temperature on each battery, or pair of batteries if they are paralleled. Absolute values, and differences in values will lead to various warnings, alarms, and fail safe actions. All of the thresholds are are adjustable so I can fine tune as I learn. It will also let me start out with very conservative operation, then expand a little bit at a time while monitoring how everything works.
- I only have two charge sources to contend with at the house, and both are programmable. I plan to program each to essentially be a two stage charger, with a bulk voltage to perform charging, followed by a float voltage down at the battery resting voltage. There will be no absorb stage, with an immediate transition from bulk to float. I expect float will require some tweaking to prevent excess battery cycling and get solar to carry the loads once batteries are charged, but while solar power is still available.
- The only fail safe I'm currently planning is a main disconnect contactor that will be controlled by the PLC. The only time the batteries will disconnect is after all alarms have been triggered, and if the situation remains uncorrected and the batteries are facing imminent damage because of over charge or under charge. This should never happen under any sort of normal operation.
- I'll also have to adjust the Automatic Generator Start (AGS) stop/start. The start voltage will be adjusted to match LFP voltages, but starting will probably have to happen more quickly once the low voltage point is reached. LFP battery voltage drops very quickly when they reach their fully discharged range, and it will be important to get the generator running and charging started before the batteries plummet. With the FLA batteries currently, I look for a low voltage level lasting for 15 minutes, or maybe even 30 minutes (I can't remember exactly). With LFP I'll probably trigger a start after 15 seconds of low voltage. It takes 45-60 seconds to get the generator on line, so that should be enough.
There must be more, but that's all I can think of right now.
Next up will be a bunch more detail on the BMS that I have built so far.
I welcome comments and suggestion?
No suggestions, just watching with interest from afar. And best of luck!ReplyDelete
Peter, good next article. I heard no mention of current monitoring capability. It would give you an ability to do a more adaptive low voltage genset start. If the system is supplying some high current loads (freezer+refer) you might opt for a shorter low voltage sense timeout. You may find other conditions where current monitoring is useful. Gas-gauge functionality is one, of course. This can give you a prediction on when the genset will be starting.ReplyDelete
I know I'm becoming something of a Cassandra with regard to security but do take a deep look at the exploitability of SCADA. It's a known easy system to corrupt and needs protection.ReplyDelete
Hey, they didn't believe Cassandra either.
Cheers buudy and carry on. Very happy for you to have this great project to tackle/enjoy.
Hi Peter, I am a fan of LFP batteries and had them operational in my vessel for 5 years now. Charging from engine alternators, victron inverter charger and recently 4kw of solar panels, they have been flawless. I chose Genasun back then and as you point out the risk is they stopped production, which they have. Hopefully I still have many cycles up my sleeve!ReplyDelete
Some more info can be found here: https://boatmags.com/lithium-ion-batteries-explained/
I am sure you will be happy with the result.
In case you haven't seen this:
The polar opposite of what you're planning, but lots of apps communicating over BT :). Seriously, still seems a lot of moving parts.
I'm surprised you're going for an 48V bank. Assuming mostly 24V consumers, this seems to necessitate an always-on inverter?
OK, first I need to apologize to everyone who has commented. I managed to do something silly, and never got any notification of these comments. I'll try to catch up and respond now....ReplyDelete
Dean said "Peter, good next article. I heard no mention of current monitoring capability. It would give you an ability to do a more adaptive low voltage genset start. If the system is supplying some high current loads (freezer+refer) you might opt for a shorter low voltage sense timeout. You may find other conditions where current monitoring is useful. Gas-gauge functionality is one, of course. This can give you a prediction on when the genset will be starting."ReplyDelete
Good point, and I might do this in the future. The issue comes into play during charge as well as during discharge. The higher the charge current, the higher the voltage will be when you reach any given charge level.
For now, I have no current monitoring, and am instead relying on the power system's inherent limits. The land-home system has max generator charge capacity of 140A which is about .4C. Worst case the solar could be at full output at the same time bringing it up to 200A or 0.55C. I'm thinking I can set voltage trip points that are "close enough" with those power limits.
On the boat this might become more important, but time will tell. The nice thing about the whole PLC approach is that if I decide to add current monitoring, it's quite straight forward to do it.
Larry said "I know I'm becoming something of a Cassandra with regard to security but do take a deep look at the exploitability of SCADA. It's a known easy system to corrupt and needs protection."ReplyDelete
OK, I had to go look up SCADA. Everything is on a LAN that is behind a NAT router. No cloud services which make me nervous with respect to hacking, and also just what the company might do with the data themselves. So other than some hack in my router... I don't know, I just can't get too excited about it. My data and I are just not that interesting.....
Greg said "Hi Peter, I am a fan of LFP batteries and had them operational in my vessel for 5 years now. Charging from engine alternators, victron inverter charger and recently 4kw of solar panels, they have been flawless. I chose Genasun back then and as you point out the risk is they stopped production, which they have. Hopefully I still have many cycles up my sleeve!ReplyDelete
Some more info can be found here: https://boatmags.com/lithium-ion-batteries-explained/"
It's great to get another positive report from an LFP user. I think the apparent complexity today will be second nature in the years to come, and we will all be doing it.....
And I haven't seen the boatmags article and am reading it now. Thanks for the link!
Anon said "In case you haven't seen this:ReplyDelete
The polar opposite of what you're planning, but lots of apps communicating over BT :). Seriously, still seems a lot of moving parts.
I'm surprised you're going for an 48V bank. Assuming mostly 24V consumers, this seems to necessitate an always-on inverter?"
Thanks for the link. More good reading....
I'm not a fan of bluetooth for anything other than a small set of non-critical uses. Even keyboard and mouse operation is problematic. For periodic config of devices I can see it, especially when the alternative is some special adapter that is always expensive and typically doesn't work without a wrestling match.
Re 48V, keep in mind that this first system is for a land-based off-grid system. That's why it's 48V. And yes, the inverter is always on. At some point you decide you just want "normal" power, and leave it on. At the house, an inverter has been on for 18 years now.
The boat will be 24V, but I did seriously consider 48V. But that's a whole other discussion. And like the house, the boat will be an inverter-on deal. Our last boat was, and so will this one. What I do look at is the standby power draw of the inverter to keep that as low as possible. It's definitely wasted power. I have resigned myself to wasting some to have AC power all the time, but there is no need to waste any more than required...
Having a similar background as you, Peter (electronics, computer science, tech exec), I also share your ideas on cruising and geeking with systems.ReplyDelete
I have a similar power system approach on our 56' sailing catamaran. 900 AH, at 12V, of CALB 180 cells, connected as 5P4S, but with the tweak of using a copper bar to connect, which ends up creating a serial connection from every cell in one parallell group to every cell in the next (avoiding potential uneven currents to/from cells within a group).
I also decided the best approach to a BMS was one that was home-stitched. I'm starting out using Raspberry Pi Zeros as the monitor/controller, as they are cheap, easily available, and exceptionally flexible to program. Whether this approach will ever migrate to a PLC architecture is too early to say.
I would caution against the approach of providing a float charge stage at your charge controller, as the LFPs really dislike being kept at a high SOC. Better to cycle the batteries (forcefully, if necessary, by eliminating all charge sources until a sufficiently low SOC has been achieved).
Also, an approach to consider would be to kick in the charge source as a load carrying supply near the bottom of the bank's SOC range. Allow enough current to prevent further discharge, but not enough to significantly increase the SOC. I'm bringing water heater elements online, as needed, to pull down SOC when too much energy is available.
For the benefit of other readers, an LFP battery is very much the antithesis of a lead-acid battery. It hates being overcharged, can't stand being kept at a high SOC for any length of time, but loves being kept near, or even at, the equivalent of "empty".
Also, I am using a single charge controller, which provides "marching orders" to all potential charging sources. No other charge logic is allowed, and I exclude any third-party products where such control is not possible.
Looking forward to reading more about your BMS and how it goes implementing your system.
It's great to hear how you have approached the same problem. Like with so many boat "things", there are lots of right answers.ReplyDelete
I considered using a Pi or Arduino, and you have shown that it's a perfectly viable approach, and certainly much less expensive than PLCs.
Re "float", my plan is to configure the chargers to support loads once the batteries get somewhere in the middlish range of charge. On one hand, I want the solar carrying loads after the batteries are fully charged, not the batteries. The batteries are there for night time and cloudy days. But on the other hand you don't want you batteries held at full charge for days and days on end.
I don't think this is a big problem in an off-grid house since the batteries are guaranteed to cycle down over night, every night. And my charging cut off point will be below 100% to begin with. I think the bigger problem is on the boat when it's left of shore power for extended periods of time. In that case, you want the shore power to carry loads, but with the batteries maintained at a mid state of charge. They key, I believe, is to set the float voltage to match that state of charge.