Regeneration on Full Battery, bad news :(

[Jason McNeil]

Energy regeneration from braking is of mixed benefit: while it is nice that some of the recaptured energy can be recovered back to the battery pack, the information provided from the battery industry is that any slight over-charge to a fully-charged cell will considerable 'stress' it, & dramatically reducing the packs  life expectancy.

 

Method: This test was on a fully charged King Song 800W, 680Wh, connected to an EagleTree eLogger with a sample rate of 50Hz. According to manufacturer, the instrument can only record current in a single direction, so instead of the typically configuration, BAT connected to the Battery & ESC to the controller, for this test, the connectors were reversed (Bat-> CB, ESC-> Battery pack), with the expected result that only power from regeneration is captured.

Result: fortunately the eLogger is able to record both the power draw during acceleration & recovered energy in braking. In the output graph output, we see that the voltage drops from 67v to below 62v in a couple second &  during hard-braking, the voltage immediately springs back, & in some cases, exceeds 72v. 

Discussion: for those using their eWheel in low-lying flat areas, these transient excesses into the +67v territory shouldn't worry to much about it. However, for owners wishing to use their eWheels on continuous descents, especially with fully charged batteries, it could do a great deal of harm to their eWheel's battery pack. Devices with less cells are more susceptible to degradation, as there is proportionally more stress per/cell (same energy, but cells to soak it up..).

In light of this evidence, it is strongly encouraged, that before engaging in any prolonged braking on a full pack, that they either a) the user not fully charge the pack,  try to use up some of the pack before heading out.

Future work: the issue of degradation on account of regeneration has not been satisfactorily addressed by any of the manufacturers—the exact degree of this problem needs further research & investigation. The ideal solution would be to have the controller rate-limit the voltage/current to the battery-pack within the battery manufacturer's defined limits, dissipating the excessive energy into some sort of thermal resistance with active cooling. These are not trivial solutions, but neither are they insuperable—I believe that many of the premature Wheel failures are primarily a direct result of over-charging during regen.      

UPDATE:

It should be stressed that over-charge phenomenon is not specific to King Song; later on I did similar set of tests with the Ninebot E: fully-charged voltage is around 61v. While testing discovered the cut-out point is 1500W, where the voltage falls through the floor at 47.24v, before cutting out. It's a curious result, because even though the KS 800W only has twice as many cells, the voltage is far more volatile under load. In these tests, my acceleration was about half as much as the KS800 before giving out... 

[KaleOsaurusRex]

This type of work can only be done with niche knowledge you posses Jason, and I thank you for taking initiative.

Would you say that the high voltage tilt back set at implemented in the King Song is enough to protect the rider? Tina claims it is implemented at 69v, so that if the rider tries to ride downhill they will need to step off of the unit and use some of the power before continuing down hill, perhaps doing a few circles, or riding uphill for a minute or so before continuing downhill.

Does that high-voltage threshold prevent the undesirable battery degradation you are describing here? It seems that it would be prudent to caution riders against riding downhill near fully charged, as even though the unit has protections implemented it still probably isn't very good for the battery if you approach that high-voltage threshold. Am I understanding your results here correctly?

[Chrisxr2]

All the routes from my house are downhill for at least 500 metres before they level out.

[Jason McNeil]

There was no evidence of 69v over-voltage tilt-back protection, then again, this was just short bursts. Probably the best thing to do at the moment, is warn new buyers about the risks of over-charging...

[Gaston]

Segways can handle this issue. 

[Tilmann]

Thank you for the research, @Jason McNeil! From what I understand, the combination of a full battery and regenerative braking is a loose/loose situation: either the BMS shuts down and sends you flying or you fry your battery - which may even set it on fire when done excessively. 

Do you already have enough insight to start quantifying the problem? Like, what amount of energy are we looking at per 1m regen braking down a steep road? I have no idea of even the magnitude. Example: would the energy produced fill a large capacitor within centimeters or could that easily buffer away the excess energy for hundreds of meters? Or: how many car headlight bulbs would it take to burn off the excess? I would guess, getting rid of 1 KW heat for more than 10 sec could become really difficult.

@Gaston: It sounds more realistic to manage such a challenge on a beast as big as a Segway. Do you know, how they do it?

[Jason McNeil]

@Tilmann, thanks, anything over 67v is a problem. Another issue that I didn't mention in the main article, is that even if the charge voltage is under 67v, there is also a Max charge current that is being exceeded; additionally, Li-ion best practice is that there should be a period of time between discharge & charge. In the case of hard acceleration/braking there is not enough time for the cells to reach internal chemistry equilibrium, apparently this is a bad thing...  

[DiasDePlaya]

The solution is simple: If you will ride downhill just start with partially charged battery. Or just start your ride from the lower side, first go uphill.

[esaj]

I don't know if short-term regeneration is that harmful (well, certainly it's at least not good) for the cells, especially with multiple packs soaking up the excess, but there should be some voltage regulation... I've been looking at the possible solutions over this week, but due to the constrained space, high power spikes etc. there seems to be no easy solution (and it doesn't help that the things happening in the motor and the half-bridges during regenerative braking aren't that simple ).

"Rheostatic" (sometimes called dynamic) braking usually shorts the motor through a power resistor or multiple resistors (which probably need a largish heat sink), but it would probably need fairly precise control to not brake "too effectively", as the balance must still be maintained. Shorting the motor without resistance seems to pretty much "slam" it to stop as fast as possible, but it's extremely straining on the motor itself too (as there's no resistance except the internal resistance of the windings etc., so the current will blow sky high), and not really a viable option for self-balancing device, unless it can be controlled precisely.

Electric cars have both regenerative and mechanical braking, and the brake controller decides which is used (or a combination of both). In case of mechanical braking, the energy is sucked by the brake discs/pads, but of course that's also not a viable option for our wheels.

Maybe the "plugging"-type electric braking could be the answer (which is basically using battery power to slow down from what I've understood), but requires changes in the driving bridges, as the polarity from the battery has to be reversed. Or regulating the voltage separately when the direction of the current changes, this again means burning off excess power in a resistor or something, but not as much, as at least part of it could still be used for charging the batteries (below 67V for example)... Or maybe we could drive another motor that's spinning flywheels inside the wheel on both sides of the motor?

 

[Gaston]

Thank you for the research, @Jason McNeil! From what I understand, the combination of a full battery and regenerative braking is a loose/loose situation: either the BMS shuts down and sends you flying or you fry your battery - which may even set it on fire when done excessively. 

Do you already have enough insight to start quantifying the problem? Like, what amount of energy are we looking at per 1m regen braking down a steep road? I have no idea of even the magnitude. Example: would the energy produced fill a large capacitor within centimeters or could that easily buffer away the excess energy for hundreds of meters? Or: how many car headlight bulbs would it take to burn off the excess? I would guess, getting rid of 1 KW heat for more than 10 sec could become really difficult.

@Gaston: It sounds more realistic to manage such a challenge on a beast as big as a Segway. Do you know, how they do it?

No, i need to check.

 

/edit

Regarding documentation: The Segway slows down. Can't believe it's all. 

[cg]

Energy regeneration from braking is of mixed benefit: while it is nice that some of the recaptured energy can be recovered back to the battery pack, the information provided from the battery industry is that any slight over-charge to a fully-charged cell will considerable 'stress' it, & dramatically reducing the packs  life expectancy.

 

Do you know how long time the batteries are stored with the charge?

 

The bad behaviour observed for instance for 4.35V, does that mean the battery is every time charged to 4.35V?

[Daan]

Just to add to this: the Solowheel with its custom BMS will start vibrating the pedals when going downhill with a full battery forcing you to get off; (vibration happens also on a low battery, or over-drawing the battery). The easiest solution is to first go uphill for a  little before going down. Also, on the Solowheel when you go really slow downhill, the wheel will not use regeneration but instead put the wheel in reverse and actually use energy; this means that the 'over charge'/vibration will not happen.

Don't know how this works on other brands.

[esaj]

Do you know how long time the batteries are stored with the charge?

The bad behaviour observed for instance for 4.35V, does that mean the battery is every time charged to 4.35V?

The image is from here, read for details:  http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

Edit: Further it refers to this:  http://www.sciencedirect.com/science/article/pii/S0378775302003051  , don't know if it's available for free somewhere.

 

[MrBump]

Seems like a lot of energy goes into keeping the rider upright.  Has anyone here got to the bottom of a hill with more lights than they did at the top?

[Mono]

Seems like a lot of energy goes into keeping the rider upright. 

How do you conclude this? So far I haven't seen much, if any, evidence that this is actually true. I am still happy to learn though, as always. 

Thank you for the research, @Jason McNeil! From what I understand, the combination of a full battery and regenerative braking is a loose/loose situation: either the BMS shuts down and sends you flying or you fry your battery - which may even set it on fire when done excessively. 

Do you already have enough insight to start quantifying the problem? Like, what amount of energy are we looking at per 1m regen braking down a steep road? I have no idea of even the magnitude.

If you can climb the very same road, it's not so difficult to make some reasoning about the order of magnitude. If you want to break with the same deceleration as you can accelerate we talk about a few hundred Watts. 

 

[Cloud]

Seems like a lot of energy goes into keeping the rider upright.  Has anyone here got to the bottom of a hill with more lights than they did at the top?

i do very often. If they battery went from , say, 6 bars to 4 bars climbing the hill, it most certainly goes back to 6 bars by the time i get to the bottom of the hill. Usually happens to me when i cross the Brooklyn Bridge on my kingsong

[MrBump]

i do very often. If they battery went from , say, 6 bars to 4 bars climbing the hill, it most certainly goes back to 6 bars by the time i get to the bottom of the hill. Usually happens to me when i cross the Brooklyn Bridge on my kingsong

So logically it uses zero energy when on the level?  I'm definitely getting a King Song.

How do you conclude this? So far I haven't seen much, if any, evidence that this is actually true. I am still happy to learn though, as always. 

If you can climb the very same road, it's not so difficult to make some reasoning about the order of magnitude. If you want to break with the same deceleration as you can accelerate we talk about a few hundred Watts. 

 

I was just taking that from what I read on here.  I don't know the numbers, and this energy may be used in a very 'spiky' way (with little energy used in riding a very flat surface and most dissipation happening in very short bursts), but it seems that hitting irregularities in road surfaces can easily overstress low-power machines, or even quite powerful machines when at low battery.

[Jason McNeil]

How do you conclude this? So far I haven't seen much, if any, evidence that this is actually true. I am still happy to learn though, as always. 

Niko's got a point, I don't believe there is any intrinsic reason why there would be a large penalty by the fact the rider is supported by the balancing system when cruising at speed. The best way to validate this is with direct comparison with eBikes: there's a number of insightful threads at endless-sphere.com. In the below graph it shows 14Wh/km @26.5kph, which is only 2Wh/km less than with the KS 800W at the same speed. The drag coefficient of the standing position is of course more than a typical ebike rider, so can assume that this slight difference can be directly accounted for by wind resistance.  
 

https://endless-sphere.com/forums/viewtopic.php?f=2&t=48191

Do you know how long time the batteries are stored with the charge?

 

The bad behaviour observed for instance for 4.35V, does that mean the battery is every time charged to 4.35V?

Battery testing protocol is usually charge to set voltage-->wait several minutes-->then discharge. Storage at overcharge may be important too, but in this example it's a cycle test. 

[MrBump]

I'm guessing this test was done on a near-perfectly smooth, level surface...

Though if overall energy use from balancing is pretty low in the general case that's quite interesting.  Riding a non-electric unicycle is much more tiring than riding a bicycle at equivalent speed over the same distance from what I hear.

[Jag_Rip]

Forgive my stupidity, but what if you have (as mentioned) a capacitor or some excessive energy consuming thingy that only activates when you generate more power than the battery is able to store? Like a hairdryer engine (stupid idea I know, but for the sake of "energy consumption")

[Jason McNeil]

It might be possible, but capacitors have pretty rubbish energy densities compared to Li-Ions & wouldn't really help for continuous descent situations. 
http://lab.bitluni.net/cap

[MrBump]

Forgive my stupidity, but what if you have (as mentioned) a capacitor or some excessive energy consuming thingy that only activates when you generate more power than the battery is able to store? Like a hairdryer engine (stupid idea I know, but for the sake of "energy consumption")

That doesn't strike me as stupid - it doesn't seem like it beyond the wit of man to have some kind of device to bleed off excess energy in these cases.  An element and a heat sink, maybe?

[esaj]

The capacitor would likely have to be in the farad-range (ie. "full" farads, not micro- or pico-farads like the ones you see in circuit boards), and those are about the size of a coke bottle. Still maybe doable, don't know?

 

[Jason McNeil]

The capacitor would likely have to be in the farad-range (ie. "full" farads, not micro- or pico-farads like the ones you see in circuit boards), and those are about the size of a coke bottle. Still maybe doable, don't know?

Yes please, I'll take a dozen of those...   

[MrBump]