Energy density comparison - Lifepo4 and LiPo

[swbluto]

So, let's choose the highest density packs of each chemistry. I don't actually know that info, but it seems Ping's are reasonably dense and so are zippy's. I've heard from Luke that Zippy's are usually two times as energy dense, so I'm wondering how true this is.

So, to calculate energy, let's take the voltage to be the average output voltage for "normal" use and multiply by AH. To calculate energy density, let's divide that by weight.

A 48V 10Ah ping pack packs about 480 wh and weighs 10.5 pounds or 4.76 kg. That assumes 3V average output voltage per cell and a nominal 10 AH discharge. The density is 480wh/4.76kg which is about 100wh/kg.

A 6s1P 5 ah pack zippy pack at http://www.hobbycity.com/hobbycity/store/uh_viewItem.asp?idProduct=8582 weighs. That pack weigh .754 kg and has an average output voltage of 6*3.6V(?)=21.6V and a nominal ah of 5, so that's about 108 Wh. The nominal density would be 108/.754 = 143 wH/kg.

Ok, so maybe people are considering the "real ratings", so let's get beyond the nominal Ah rating and go with "real" ratings, right after the "break in" period. It seems people with balanced ping packs usually have 10.5-11Ah. People with zippy packs have AHs of 6? Ping's density would then be (Assuming 10.5 Ah) 106 wh/kg and Zippy's would be 171 Wh/kg. That's a difference of 171/106 = 1.61 times. Where's the 2x figure coming from?

I guess if you took a 10Ah ping and expected it to continuously discharge at 4C, then the output voltage would be so low that the output energy density would probably be twice or more less than that of a zippy pack. But for usage within its ratings?

Despite that, an energy density of 1.5-1.6 times is still appreciable. That's about the difference between 20 pounds and 30 pounds or the difference between 50 and 75 lbs.

 

[Doctorbass]

I have great REAL data abaout Ping battery.

Since I got 7 Ping battery for my familly's ebikes, I tested every of them to ensure everything is ok ( Doc QC test..)

The battery are the new V2.5 48V 15Ah made of 48 cells ( 16s 3p(5Ah))

I tested them with resistive load with an average current of 20A with avg power of 1000Watts ( two large resistor with a 8" fan)

Every of the 7 battery was fully charged until every 16 leds of the BMS was active and had last for 5 full hour.
( Final voltage 1 min after discnnected from charger 58.8-59.1V )

The BMS really cutted the battery at 40V.

At 1000W ( 1.33C)
-the max i got was 772Wh
-the min i got was 691Wh

so let's say (772+691)/2= 732Wh

and the battery weight:7.5kg

it's 732/7.5 =

AVG PING BATTERY DENSITY over 7 battery tested at 1.33C (1000Watts)
97.6Wh/kg

so it confirm what you calculated with real life use of 1000W continuous

but.. is it really mass density over volume that is really important for ebikes...

I almost think it would be volumetric energy density..

for ebikes.. the weight is not really a problem compare to the room to place battery that is missing on every frame to get a great center of gravity.

zippy seems very interesting but i still dont like the fire risk

Doc

Doc

 

[Toorbough ULL-Zeveigh]

Watt-hours/litre = energy density.
Watt-hours/kg = specific energy or in a pinch, gravimetric density.

i'm also of the opinion that it's the available real-estate which is at a premium on a bicycle.
that's why for me even tho the lifespan & cost of ferro-phosphate variants is attractive, in my view their larger bulk is actually a step backward with a lower energy density (volumetric) than even NiMH.

that may be where the 2x specific energy fig is coming from.
thumbnail specific energy calculation i've been doing of what's available has been fairly consistant.
roughly speaking NiMH tops out 80 Wh/kg, LiPo around 160+ Wh/kg with LiFe falling in between from 100-110.

 

[swbluto]

I just weighed my battery and it appears it's about 11.5 lbs. So the nominal wh/kg is about 93, while the "realistic" is around 98 which corroborates Doctorbass's results.

It seems like it'd be interesting to measure the volumetric energy density. Let's get some numbers on that.

My preassembled 24v10ah ping pack is (Not including the BMS or output wires)

13.5 cm x 14.5 cm x 14.8 cm which is 2897 cu. cm.

The nominal volumetric energy density is then 480 wh / 2897 cu. cm = .165 wh/cm.^3.

Using the same pack I linked to earlier, its stated dimensions are:

145x50x48mm

Which is 14.5 cm x 5 cm x 4.8 cm or 348 cm.^3. It's nominal watt hours, as deduced earlier, was 108 wh which makes the volumetric energy density 108 wh / 348 cm^3 = .31 wh/cm.^3.

.31/.165 = 1.87 times as much.

If these figures are to be believed, the volumetric energy density of this kind of Li-poly is almost 2x that of Ping and at least 2x that of Ping in practice shortly after each's break-in period.

Wow, this is impressive. 1.7x the gravimetric energy density and almost 2x the volumetric energy density. Ok, so the talk about lipoly having twice the energy density of ping is definitely supported by the numbers.

Kind of makes we want some lipoly now. I can get an awesome scooter with its limited room belly tray with a high top speed and pretty good range, in theory. In practice, I don't want to put the Lipoly so close to the ground and I don't want to sacrifice its wonderful complement to bus transport.

 

[swbluto]

One factor I haven't liked about lipoly batteries is cycle life. The upfront cost may be cheap for the power levels you're getting, but the longevity doesn't exactly bode well for the long-term assuming you use the batteries often.

But, I then had a thought, how does this cycle life compare to LiFePO4?

I've heard under "normal conditions" (charge to 100% and discharge to 20%) that the cycle life is usually around 200-300 to 85% of initial charge capacity. With charging to 80-90% and attentive storage practices (For longer term storage, cool it as much as possible at a 20-60% DOD), this could be doubled. How does that compare to a lifepo4 battery?

A lifepo4 battery would have a lifetime around 1000x at or near its maximum C rate (Which I obligingly do with my ping batteries - I don't even come close with my A123s). Ok, so we're looking at a lifespan of nearly 2 - 4x as much.

But, what about "energy density" lifespans? Sure, you may have got to 85% of your lipo's initial capacity, but its energy density is *still* higher than that of a LiFePO4 battery so if a LiFePO4 system was usable with your design, then the lipo still has some "energy density" life left. How much left? I don't know. I haven't seen the capacity discharge/charge cycle life graph for LiPo, but I have seen it for various lithium chemistries and unlike SLA, it seems to be pretty linear (SLA is more parabolic). That basically means that it makes calculations easier.

So, if the volumetric energy density of lipoly is twice that of equivalent lifepo4, then it'll need to be cycled to 50% of its original discharge before it has lifepo4 equivalent volumetric energy density. Assuming 300 cycles to lose 15%, that means it would take a little more than the 900 cycles to lose 50%. But, after 900 cycles, lifepo4's own capacity has declined to 85-90%. So, to make it truly half, it needs to run somewhere around 1000 to 1100 cycles to have lifepo4 equivalent volumetric energy density. Not bad! It performs far better while lasting about the same amount of time to get to lifepo4's volumetric energy density. I think it might also be cheaper, but my mind seems to be getting a little confused by the initial conditions involved.

For the gravimetric energy density, it sounds like it takes somewhere around 600-900 cycles to match lifepo4's. Still not bad.

If were talking about powerful lifepo4 like a123 round cells, you also have non-100% packing ratios which further decreases a lifepo4 pack's volumetric and gravimetric energy density which even makes lifepo4 worse. But I've heard a123 is coming out with prismatic/flatpacks, so maybe we'll get better specimens to compare.

So, all in all, lipoly is looking pretty rosey.

I still have to wonder about the long-term costs and possible long-term cost strategies. I don't really know if it's cheaper or more expensive with intelligent buying strategies. But if it is more expensive, it seems to be by a small percentage amount, in which case the extra performance seems to be by far worth it for high powered setups.

Of course, this is speculation that is ultimately subject to hard data. I would be extremely interested in some LiPo cycle life tests.

 

[Doctorbass]

Another IMPORTANT parameter us teh voltage over the discharge curve!

A LiFePO4 is pretty flat util 10% SOC... but a lipo is not so flat... the voltage curve drop faster.. so the last Ah are not so powerfull ( V x I ) at 90% soc and ( V x I ) at 20% soc


LiFePO4 are have more flat discharge and this is important i think.
Doc

 

[liveforphysics]

LiFePO4 often looks flat, or even increases voltage during discharge because it's crap Ri makes it start heating up. lol

LiCo LiPo looks like a perfect voltage discharge curve, because the cell isn't even getting warm as it discharges.



With respect to cycle life, many cell spec sheets for LiPoly cells show an extension of cycle life by a factor of 5x or even 10x by simply charging to 4.125v/cell rather than 4.20v/cell. Some RC guys push cells up to 4.3, or even 4.35v/cell for races and things, and just accept that the pack will only be good for a few cycles. With my E-bike, I charge to 4.125v/cell, and when i occasionally check on how the cells are doing, I'm so far still breaking in and making gains in capacity... As far as the loss of energy storage from going to 4.125 vs 4.2, the HC cells are often so under-rated, I still get over 5Ah from only charging to 4.125v. If I charge clear to 4.2 or 4.25v, I get 5.5 to even 6Ah. I think Methy said he has some "5Ah" cells that he normally cycles at 6Ah.

 

[swbluto]

Another IMPORTANT parameter us teh voltage over the discharge curve!

A LiFePO4 is pretty flat util 10% SOC... but a lipo is not so flat... the voltage curve drop faster.. so the last Ah are not so powerfull ( V x I ) at 90% soc and ( V x I ) at 20% soc


LiFePO4 are have more flat discharge and this is important i think.
Doc

That seems to be a bit of a psychological detriment, and that could be addressed by several means.

As far as top speed goes, just ensuring that the last tail is high enough to support the speed you want and the extra at the beginning is a bonus. Also, set the acceleration bar for the end performance and it won't be similarly disappointing. But, expectations tend to be heavily influenced by prior performance, so that may be hard to do in practice given human's natural psychology.

Also, it can be argued that the voltage drop is a benefit. With it, one could tell when their batteries are getting low and how low they are. It could also be used to semi-accurately gauge how much battery capacity you have left, which isn't true of lifepo4. Does it have 30% or 80% left? You won't find out until the middle of the ride with LiFePO4. LiPo? Instant know.

People say that not having an LVC on a123s is "suicidal" because you'll inevitably kill the pack. Poppy cock I say! It helps that I have experience with lifepo4 voltage drops when I killed my first ping pack, maybe, but I caught my a123s performance drop when their resting voltages were at 2.73-2.77 volts (They were extraordinarily balanced even then). But if you don't have that experience, you might actually kill a couple of cells the first time around. And maybe second time if your financial loss wasn't great enough the first time.

 

[swbluto]

With respect to cycle life, many cell spec sheets for LiPoly cells show an extension of cycle life by a factor of 5x or even 10x by simply charging to 4.125v/cell rather than 4.20v/cell.

I seem to have difficulty finding data sheets with that claim. However, I did see that motorola suggested 4.1 volts was the optimum charging voltage for cycle life for li-ion.

And I found this resource on LiPo testing and it's very informative. http://www.rcuniverse.com/forum/m_3559995/tm.htm

It turns out that the capacity doesn't just gracefully decline to zero percent. It declines to somewhere between 50 and 70% (Didn't find out the exact number), and then it plummets due to dendritic salt formation or something like that. EDIT: correction. According to my famous reading comprehension, hardy har har,

Cycle life as defined by international standard 80% of base capacity was 450 cycles. The pack was run to complete death just to see what would happen. Results are interesting: Capacity declined slowly and steadily to 475 cycles, then plummeted. Dissassembly of cells confirms the rapid formation of salts (dendrites) as the cell dies.

That implies that somewhere in the 75-80% region = death.

So it seems the cycle life is limited. What's the actual limit to useful capacity? With careful charging and low operating temperatures, maybe somewhere in the 500 to 800 cycle region. I don't know. I kind of wonder if this "sudden cycle death" pertains to LiFePO4?

I found information on normal Li-ion at the battery university - http://www.batteryuniversity.com/partone-12.htm

Apparently a charge voltage upto 3.92 volts can double the cycle life as compared to a standard 4.2V protocol for cobalt. Given the nature of how cycle life decreases as charging voltage increases, it seems that 4.125V would increase cycle life between 1 and 2 times and 1.5 times sounds like a close guess. But I don't know if the 3.92 volt remark refers to the topping voltage or the voltage to begin "topping".

(That was the most quantitative information on there about cycle life increase by lower charging voltages.)

Ok, yes, I finally found a quantitative source on charge voltage and cycle life! Even better, it has a graph!

Power Electronic's Li-ion guide

It's figure 2. it shows that the cycles double at 4.1 terminating charge voltage compared to 4.2 volts, in exchange for a decrease of 15-20%(?) of usable capacity per cycle.

 

[swbluto]

EDIT: UTTER GARBAGE

Ok, I'm going to pull out the math! I want to find out which tapering voltage maximizes the total usable capacity over the entire lifespan of the cell. As the charge voltage decreases, the cycle life increases but yet the capacity available drops. At what point is the cycle life increase not enough for the drop in usable capacity and the total usable lifespan capacity is maximized?

I'll do two linear fits on the capacity and cycle life curves and maximize the total lifespan capacity. But first, some basic formulas.



Total usable lifespan capacity(T) = caPacity(P) * cycle-life(C) -> T = P * C

V is the topping charge / float voltage.

Here are the two linear fits. 4.0 < V < 4.2 cuz it looks most linear there.

P(4.0) = 500; P(4.2) = 1250

P(V) = 3750(V-4.0)+500;

C(4.0) = 1750; C(4.2) = 500;

C(V) = 1750 - 6250(V-4.0)

T(V) = P(V)*C(V) = -3.87875*10^8 + 1.90938*10^8 V - 23437500 V^2

dT/dV = 1.90938*10^8 - 46875000 V

setting dT/dV to 0 and solving for V,

V = 4.07333 volts.

It looks like 4.1 to 4.125 volts is a pretty justifiable taper voltage.

And yes, blah blah, it depends on so many more variables than taper voltage, blah blah.

Let's calculate how much more total lifespan capacity you'll get at 4.0733 V and 4.125 volts. I suspect it'll be a negligible amount but it's nice to just to check the magnitudes.

Ratio = T(4.07333)/T(4.125) = 1.06606.

 

[swbluto]

I was tired last night and so I was prone to errors, like seeing that P was measured in mAh instead of percentage. I just made some new calculations that show 3.9 volts maximizes total lifespan capacity, but I just realized you can't calculate T = P * C at P, it's the average percentage capacity during the cell's life that matters. It actually seems more complicated than that as it doesn't seem like I can infer the capacity from figure 2. That's only the percentage of total charge available at a given cycle, and gives no information about the actual charge available at a given cycle.

Since I'm tired of doing these calculations, and I've already made a calculation that suggests 4.0 is better than anything higher than it, I just might try to force fit 4.0 and 4.125 volts into whatever average capacity / cycle formula I can find. Until then...


Anyways, as far as practicality goes, it seems that 4.1-4.125 volts might be a good compromise. You might actually see 600-700 cycles before the calendar life starts to take a hold in normal e-bike usage while still having relatively significant capacity to use in a given cycle. 4.0 volts *might* extend your total usable lifespan capacity by ~40% if cycled within a short time frame, but given 1000-1500 cycles might take an e-Bike an extra 1 - 2.5 years to complete, the calendar life deterioration may start to erode away the lifespan capacity advantages while having a per-cycle capacity 70% disadvantage (As opposed to 85% at 4.125 volts).

 

[swbluto]

I was interested in lipo's comparison with gasoline.

So, I'm using gas's values from http://hypertextbook.com/facts/2003/ArthurGolnik.shtml.

First, by weight.

Gas is 12.7 kWh / kg. Assuming 20% engine efficiency, that's 2540 wh / kg. Zippy lipo was earlier calculated to be 171 wh / kg. That's a difference of 2540 / 171 = 14.85 times less weight per unit energy output.

Gas is 8.76 kWh / liter or 8.76 watt-hour per cubic centimeter. After 20% engine efficiency, that's 1.74 wh / cm^3. Lipo was earlier calculated to be .31 wh / cm^3. That's a difference of 1.74 / .31 = 5.6 times less volume per unit energy output.

So, it seems battery technology is not that far from gasoline in terms of effective volumetric energy density. It is, however, far heavier.

 

[liveforphysics]

Do you think it's fair to compare energy from it's volume and weight as stored, but measured by it's mechanical output, but measure batteries from just the chemical storage? And if you're going to compare the best of batteries, wouldn't you want to compare to the best of gasoline engines as well? 33-35% chemical potential to mechanical can happen even with race engines, diesels can hit >40%. The best diesels are >54% efficient (but low speed monsters for ships and power gen).

 

[swbluto]

Do you think it's fair to compare energy from it's volume and weight as stored, but measured by it's mechanical output, but measure batteries from just the chemical storage? And if you're going to compare the best of batteries, wouldn't you want to compare to the best of gasoline engines as well? 33-35% chemical potential to mechanical can happen even with race engines, diesels can hit >40%. The best diesels are >54% efficient (but low speed monsters for ships and power gen).

Yes, I knew there was going to be critiques based on variances in the parameters. I thought I would compare a typical car engine efficiency with battery specs. Even though, that's probably off by 20% because I didn't take into account electric motor efficiencies but, hey, the best of lipo gets 10-20% more than that so maybe it's not far off.

Btw, I was looking at small gas engines and noticed that it stated a fuel consumption or something efficiency related of "340 g/kWh". Do you know what g stands for? I'm wondering what efficiency would that be if I were to operate that engine at peak efficiency, but I kind of need to know what g is to start the calcs.

Oh wait, a quick google search suggests that means grams. Now to calculate...

Just some quick notes for my calculations...

12.7 kWh / 1000 grams.

340 grams / kwh.

That means 1000 grams puts out 1000 * 1/340 = 2.94 kWh.

2.94 kWh / 12.7 = 23%.

That suggests the small 50cc 4 stroke and below honda engines are 23% at peak efficiency. I wonder what bigger honda 4 stroke engines get?

Another source says that the 120cc honda 4-stroke engine gets 313g/kWh. That's (340/313)*2.94/12.7 = 25.1%.

 

[liveforphysics]

The best engines do 155g/kwhr.

These little tiny engines you're comparing are kinda the worst out there for efficiency.

 

[swbluto]

These little tiny engines you're comparing are kinda the worst out there for efficiency.

Interesting. I'm guessing this is probably due to displacement? I've noticed that smaller car engines tend to result in a higher MPG which is one of the reasons why the older insight gets the highway MPG that it gets (Along with an awesome effective drag area for a car), so I was under the impression that for a given application, smaller = more fuel efficient? It's probably not that simple...

Is there a specific type of small engine (for a 30 mph bicycle) that gets the best fuel-consumption? Like... a twin valve, computer-injection, 12:1 displacement ratio type? I'm half and half just drawing words at random. Or do they all suck at that scale?