Typical Cell Operating Limits

Manufacturers ratings for LiFePO₄ battery cells have become more conservative in recent years as more experience was gained with the practical operation of these cells. Nowadays (2017), the typical operating specifications for LiFePO₄ prismatic cells look as follow:

CALB SE series cells

Maximum Charging Current 200A
Maximum Constant Discharging Current 200A
Nominal voltage 3.2v 4 cell bank = 12.8 v
Charging Cut-off Voltage 3.65 V 4 cell bank = 14.6 v
Discharging Cut-off Voltage 2.5 V 4 cell bank = 10 v
SOC Usage Window 10% ～ 90%
Charging Temperature 0℃ ～ 45℃
Discharging Temperature -20℃ ～ 55℃
Storage Temperature within 1 month-20℃ ～ 45℃
Storage Temperature within 1 year -20℃ ～ 20℃

Assume a 400 A/hr house bank – this would give a usable capacity of around 360A/hr max.

 

Balancing:

LiFePo4 cells are all slightly different.  Even well matched cells with consecutive serial numbers are each a little different from each other. This means that they will each have a slightly different capacity, and one cell will always be the weakest to a minor degree. The degree is not particularly important as we never drain a battery to 100% discharge, but it does cause a potential problem if a cell has a markedly different state of charge to the others. If a single cell reaches 100% charge before the rest of the bank, it will be overcharged and damaged as the rest to the bank come up to full. Similarly, if a single cell reaches zero, while the rest of the pack is discharging, the cell can suffer reverse flow and will be irreparably damaged almost immediately. in order to prevent this from happening, packs of cells should be balanced in a way suitable to their load profile. Cells can be bottom or top balanced, and each suits a different use profile.
For heavy discharge use such as an electric vehicle or drive pack, bottom balancing is the safest, as this means the whole pack reaches empty at the same time. Charging can follow the lowest capacity cell in the stack and the pack will remain safe.  For our house bank, which is not subject to very high discharge currents and ultra deep discharges, top balancing is better. In this method, all of the cells should reach maximum capacity at the same time, although care has to be taken not to discharge the pack lower than the cut off voltage for a single cell.

Bottom balancing is time consuming as each cell needs to be carefully discharged to it’s lowest safe state prior to assembly of the pack. Discharge rates should not exceed 0.1c during balancing which means that with multiple high capacity cells, this can take a considerable time (10+ hours to discharge each cell). The pack is then assembled and charged until a single cell reaches the full state. In general bottom balancing is of most use where battery packs are expected to be very deeply discharged – such as in electric vehicles or electric drive systems. It is not generally a good policy for a domestic house bank which is unlikely to be so deeply discharged.

In Top Balancing, we do not have to discharge the cells, as we are primarily interested in the fully charged state. Each cell is carefully charged to 100% and the pack assembled. Conventional wisdom (and common sense) dictates that in either case, it is necessary to actively monitor each cell’s voltage in order to control both charge and discharge cut off . Some Battery Management Systems (BMS) advocate using so called ‘vampire boards’ in order to try to prevent over charging a single cell. However given that LiFePo4 cells do not appreciably drift in relation to each other, I do not consider these of value. Providing the battery control circuitry is capable of providing both charge and discharge control / cutoff, batteries should not drift significantly from their balanced condition. Top balancing is aimed at ensuring all cells reach a fully charged condition at the same time. While it is true they will drift apart at the extreme bottom end of the charge, a house bank is unlikely to see such deep discharge conditions.

 

Charging:

Charging LifePo4 batteries is completely different to Lead Acid systems. LA is tolerant of both over voltage during bulk charge and float at significant voltage without issue, while LifePo4 does not appreciate either, and can fail when subjected to LA charge patterns. The ideal charge profile for LiFePo4 is constant current- Constant voltage. That means a constant current until the battery voltage reaches around 14.4v, then constant voltage at 14.4v until the charge current reduces to a nominal figure, (approx 0.01c = 4a).  Ideally all charge should stop completely at this point as the battery is then full. Unfortunately the standard alternator regulator in use everywhere is a constant voltage device. This will attempt to push power until it reaches the voltage set point. This means that the alternators would be running far in excess of their rated load for most of the time –  a guarantee of demise is short order.Ideally, all of the charge sources would have matching LiFePo4 control systems, ie: constant current – constant voltage with a hard limit. Ideally all charge to the LifePo4 batteries should be disconnected once they approach fully charged, however sudden disconnection of alternators using the bank high charge disconnect controller would potentially destroy the alternators due to back EMF. Before an alternator can be disconnected while running, it’s field control must be shut down to prevent excessive back EMF voltages destroying the diodes. This essentially means that engine driven alternators for LifePo4 should be actively controlled as part of the battery charge control regime.  If using LA Engine start batteries, this disconnection becomes less of an issue provided such batteries are on the alternator side of the circuit, thus never disconnected from the charge line, in which case disconnecting the LifePo4 using the high charge disconnect would not disconnect all load from the alternator, and thus would prevent the high EMF voltage spikes. It should however be noted that the charge regimes for LifePO4 and LA are different and if operating correctly for LifePO4, are sufficient to cook LA batteries as they never reduce to float level. By the same token LA float voltages of 13.8v are insufficient to fully charge a LifePo4 battery. Remember LA like to be floated, LifePo4 hate it!

 

A yacht generally contains a number of different charge sources. In our case this is expected to include, 2x engine alternators – lets say 130 Amps total, 400 watts of solar power (say 28 amps peak), 350 watts of wind (around 25 amps) and a shore power DC charger (30+ amps). Thus in theory, we can shove a maximum combined charge rate of around 213 Amps into our house bank. In reality, we are never likely to see more than around 150 amps peak. Given a typical LifePo4 house bank of around 400A/hr, and a maximum discharge of 80%, we can expect the bank to recharge in a little under 2 1/4 hours although 2 -3 hours under motor only is more likely. This is still vastly faster than recovering a similar LA bank. Charging time could of course be reduced by using larger capacity alternators, but at additional cost and load on the engines. To date I have been unable to find a comprehensive LifePo4 charge controller capable fo both monitoring correctly and controlling multiple charge sources – looks like I will have to develop one to do what I want correctly.

It would be possible to use an all LifePo4 solution for the boat, including engine start loads, but in that case, active control of all charge sources must be from the BMS in order to manage charge rate and charge stop in a controlled manner. This means the BMS becomes more complex and far more essential and is something I will investigate at a later time. Having LA in the mix does offer some degree of redundancy in the event of a battery or BMS failure. While electronics are very reliable, Catamarans are subject to more lightning strikes than monos. If a  strike takes out all of the electronics, then purely manual / mechanical systems can keep going.

Given the figures above, a charge profile needs to be constructed which will allow the alternators to produce no more than 14.4v irrespective of current load. However, in reality, voltage control would rapidly cook the alternators if allowed to attempt simple voltage limiting operation at high charge levels. Thus we need to limit current in the early part of the charge cycle in order to protect the alternator.

My catamaran is fitted with a pair of Yanmar 2GM20 motors rated at 18Hp each. As standard these are fitted with a Hitachi LR135 (35amp) alternator as standard . undoubtedly these sill need to be replaced with something more sensible. I am thinking at least 55 – 70 amps per motor would be sufficient, but would need decent control for the LiFePo4 house bank. 35 Amps per motor would equal around 0.35c charge rate at best, with massive load on the alternators – not good. Better would be an alternator capacity of around 1c, or around 100A per motor. Still needs decent regulation but offers faster recharge capability and longer alternator life expectancy. The down side is that is that it is quite difficult to drive a 100A alternator with a single belt, resulting in high belt loads and limited life. perhaps a compromise around 75Apms / motor would be more appropriate.
On the other hand, the current 35 amp alternators can manage around 0.3c so a full charge (from flat) should take no more than around 3 hours of motoring – not too bad all things considered

 

The BMS:

I have yet to see a BMS which actually does what I want to do. Generally they are concerned with the battery it’s self not the charge sources – ie: they do not address the platform as a whole.  Given that, it looks like I must develop my own. Fortunately I think this is  well within the capabilities of a number of self contained micro controllers, so should not be too difficult.

Requirements:

Charge / discharge current measurement +- 200A range
Alternator current – Alt 1
Alternator current – Alt 2
Battery Voltage measurement ( 2 decimal places, accurate to 10mV max)
Battery Cell Voltage Measurement – for failure / alert warning. (2 decimal places, accurate to 10mV)
Battery Temperature
Alt 1 Temperature
Alt 2 Temperature
Field control output – alternator 1
Field control output – alternator 2

Solar charge control set point (MPPT control)
Wind gen voltage measurement ( 2 decimal places, accurate to 10mV max)
Wind O/P Current
Wind dump control
AC charger Voltage ( 2 decimal places, accurate to 10mV max)
AC Charger Current
AC Charger Control.

High volt / charge disconnect
Low volt Load Disconnect
Alerting
Charge state monitoring and reporting.