Analysis of regen on an ebike

Get all your technical information about electric bikes here.

Analysis of regen on an ebike

Postby justin_le » Mon Dec 15, 2008 2:11 am

This June I gave a presentation at the monthly VEVA meeting ( http://www.veva.bc.ca/events/pastevents.htm ) about among other things regen braking on electric bicycles. This was before the cross-canada ebike trip but after doing some quantitative tests riding around Vancouver with various early prototypes of the regen motor controller and a datalogging Cycle Analyst. I don't think anyone from this forum was in the audience so I'll summarize the results here as well for people to have something to read over the holidays :) .

Is it Worth It?

Everyone knows that the "does the it charge the battery while you pedal" is the most common question you get with an ebike, and every ebike vendor has to deal the regen question all the time as well, and since most ebikes don't do regen there have been a lot of statements and excuses spread around justifying NOT having regen. The main one we see is that it's hardly worth all the extra cost and complexity of a regen circuit just to get a few percent back in the battery.

To set the record straight on this point, here is what a regular non-regen brushless motor controller looks like, compared to a regen motor controller on the right.

Controllers, Regen and Not.gif
Controllers, Regen and Not.gif (14.52 KiB) Viewed 8594 times


With brushless motors, the basic circuitry behind a regen and a non-regen motor controller are identical. The power components like capacitors, mosfets, mosfet drivers and such are exactly the same, and these are the items that determine the cost of and complexity of the controller. The only thing that makes a regen controller different is the manner in which the mosfets are driven. In a non-regen motor controller, the current during the PWM off cycle typically flows through the diode of the low side mofet bridge doing PWM, while in a regen motor controller this current flows through the low side mosfet which is turned ON for the OFF portion of the PWM cycle.

Current Flow.gif
Current Flow.gif (24.2 KiB) Viewed 8572 times


Anyways, even without understanding the details the important thing to appreciate is that with brushless motors, having regen does not inherently increase either the cost or the complexity of the controller, but as we'll see soon it does require significantly more sophisticated firmware.
Big Dummy with prototype 26" Grin thru-axle front hub, 40A field oriented controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled
User avatar
justin_le
100 kW
100 kW
 
Posts: 1106
Joined: Mon May 28, 2007 3:27 am
Location: Vancouver

The percentage recaptured

Postby justin_le » Mon Dec 15, 2008 3:37 am

There is also endless debate about how much you actually get back with regen, some poopooing it at just 1-2%, and other regen proponents touting the 10-20% claims. The fact is both these extremes and everything in between are true. It depends entirely on your riding environment; how hilly is your terrain and how much stop and go you have to endure.

This can be easily seen in the cross-canada ebike stats. Riding from Calgary to Northern Ontario across the Canadian Prairies I rarely hit 1% regen. It's just all flat or rolling hills and pretty much no reason to brake. In fact, it surprised me that I got as much back as I did from just the couple times I would come to a stop for a drink or pit-stop and the likes. Then if you look at the numbers through the mountainous parts of BC I got nearly 18% regen down Manning Park and 12.6% over Roger's Pass.

However, what is missing from these stats and is most relevant to the discussion are the regen figures while riding around inside the cities, since that's where most people do their biking. Here, the numbers are more consistent and ranged from 8% to 14% (which I got from riding around the bike paths in Calgary, moving slow with lots of stopping and short steep hills).

This correlated pretty well with tests in Vancouver before leaving on the trip. In preparation for the VEVA talk I took a mountain bike outfitted with a 500 watt Golden hub motor and a regen motor controller set to a 20A current limit, and then rode it along what I considered a typical stretch of road in Vancouver, straight along Main Street from Terminal Avenue to Marine Drive:
http://tinyurl.com/67psys
Route.gif
Route.gif (90.69 KiB) Viewed 8549 times


This route has a total span of just over 7km and has a net elevation gain and drop of 77 meters, or 250 feet, and the start and end point are at almost the exact same altitude. My procedure for riding and using the regen was as follows: I used the regen brake whenever I knew I needed to stop, such as approaching a stop sign or red light, and whenever going downhill steep enough that without the brake I would be coasting over 40 kph. I didn't consciously try to maximize the regen (such as by pedaling while ebraking) but at the same time I didn't engage in the momentum saving practices of a lot of regular cyclists (like cruising through stop signs, or gradually coasting to a an anticipated stop well in advance to save on pedal energy and brake pad wear).

The data was logged at 1 second intervals with a Cycle Analyst, and is plotted for both the north and south trips along with the calculated statistics. The pink line is the speed, blue line is my current draw from the battery (a 36V 8Ah NiCad), and you can see pretty clearly going to negative 10 amps or so before each stop, and then a prolonged region between -5 to -10 amps just at the end of the trip where I was going down a steeper part of the street and had to use the regen to keep the speed at 40 kph.

South Stats.gif
South Stats.gif (12.29 KiB) Viewed 8535 times


And similarly, on the return trip from Marine drive to Terminal.

North Stats.gif
North Stats.gif (11.2 KiB) Viewed 8529 times


In this case, I had better luck with the intersections and only had to stop for red lights 7 times rather than 10 times, and you can see that the % energy recaptured was slightly less (9.7% vs 11%). So, the analysis here is basically 10% regen on this route. I haven't summarized it in the graphs above, but of that regen energy 56% came from the stopping, while 44% was from governing the speed on the downhill. In both cases I used slightly under 10 Wh/km which is the typical figure we see for ebikes that travel in the 35-40 kph speed range.

So, although this is a just from a single run, you could easily conclude that we'd expect about 5% regen for city riding in mostly flat terrain, 10% regen for city riding with lots of moderate hills, and probably more like 12-14% regen if there are really steep hills.

Having since ridden all around Vancouver many times with the cross-canada regen ebike, it seems I'm almost always pegged somwhere bewteen 8 to 11% regen by the time I complete a return trip, but I probably use the regen more aggressively than I would otherwise use a mechanical brake, so this may skew the results somewhat. I would expect others with lighter bicycles would probably average more in the 6-10% territory here.
Big Dummy with prototype 26" Grin thru-axle front hub, 40A field oriented controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled
User avatar
justin_le
100 kW
100 kW
 
Posts: 1106
Joined: Mon May 28, 2007 3:27 am
Location: Vancouver

Compared to the rolling drag torque

Postby justin_le » Mon Dec 15, 2008 4:38 am

Unless you have a clutch system (often talked about but seldom if ever implemented), an ebike that does regen will always have the motor engaged, so that even if you are not using the motor it is still forced to turn causing extra drag on the system. A claim is often made that because of this you are way better off with a system that properly freewheels and doesn't have regen than one that is always engaged with with regen. The implication here would be that whatever energy is gained back through regen is less than the energy that is wasted by having to turn the motor over when you aren't using it and just want to coast.

So lets have a look at how this plays out with the trip up and down Main street. Here I was using a Golden Island hub motor. The drag torque just to rotate the hub was measured by spinning the motor at various RPMs with the axle supported by ball bearings and a torque arm/load cell arrangement used to measure the actual reaction torque on the stator. The graph below shows the results for this hub:

Drag Torques.gif
Drag Torques.gif (5.02 KiB) Viewed 8505 times


At the speed I was traveling, the drag of the hub varied between 0.4 to 0.5 Newton-meters, but for the sake of calculations lets assume 0.5 N-m throughout. So, over the course of the 7km trip, the energy expended just to turn the motor wheel is (0.5 N-m / 0.33m * 7000m = 10600 Joules) or 2.9 watt-hours.

By comparison, the energy recaptured and available to use again because of regen( regen Ah * 36V ) is 6.5 Wh and on the South trip and 6.2 Wh on the return. So in both cases over twice the energy was recaptured and than was lost by the power required to turn the wheel.

What that means is that if you ran the hub motor so that it draws just enough power to overcome its internal drag so that it perfectly simulates a freewheeling setup, you would still end up with a significant net gain over a having no motor or regen at all, putting a net amount of energy into the battery pack and no extra drag on the bicycle

Now a freewheeling setup still has to overcome the motor drag when the motor is running, so the actual results in a realistic comparison are significantly more favorable than this. If for instance the motor was being used 50% of the time, then we would have gained over 4 times more energy from regen than we could have saved from having a freewheel instead. But even for people who only use the motor on the occasional hill, they have more to gain in terms of total energy consumption by having an always engaged motor and regen than they would save from perfect freewheeling.

There are important psychological reasons why freewheeling would be preferred for a lot of people in this case, but the energy argument doesn't seem to hold.
Big Dummy with prototype 26" Grin thru-axle front hub, 40A field oriented controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled
User avatar
justin_le
100 kW
100 kW
 
Posts: 1106
Joined: Mon May 28, 2007 3:27 am
Location: Vancouver

Re: Analysis of regen on an ebike

Postby paultrafalgar » Mon Dec 15, 2008 4:52 am

Guilty as charged, Justin :oops: (for misleading the public re regen) here:
viewtopic.php?f=2&t=5047&p=75891&hilit=regen#p75891
Thanks for the highly cogent correction! :D
If we knew what it was we were doing, it would not be called research, would it? Albert Einstein
A paranoid is someone who has SOME idea of what's going on. Allen Ginsberg(?)
If the greatest pleasure is giving, be selfish - give pleasure.
User avatar
paultrafalgar
100 kW
100 kW
 
Posts: 1150
Joined: Fri Feb 15, 2008 5:41 am
Location: Forest of Dean, UK

Regen Efficiency

Postby justin_le » Mon Dec 15, 2008 5:34 am

One thing that is slightly counter-intuitive with regen is that applying more regen braking force does not necessarily increase the regen current going back into the battery pack. This is easy to understand when you realize that the maximum regen braking torque occurs when all the motor phases are simply shorted together. In this situation, the regen current simply circulates through the windings of the hub motor generating heat, and none of that energy ever flows back into the battery pack. At the other end, when the duty cycle (D) of the motor controller is such that the effective voltage going to the hub motor (Vbatt * D) exactly matches the back emf voltage of the motor, then there is also no current flowing into or out of the battery as well, though in this case there is no braking torque either.

Regen Amperage.gif
Regen Amperage.gif (15.81 KiB) Viewed 8495 times


In between these two points the current flowing into the battery follows a parabolic shape, with the maximum regen battery current occurring exactly in the middle. This is the point of maximum power transfer of energy back into the battery pack, and the hub is generating exactly half the braking torque that it would produce if all the windings were shorted. That's where you would want to be to get maximum amperage into the battery pack, but the efficiency of turning kinetic energy into useable battery energy is only 50% at this point.

In general, the less the regen braking current the higher the efficiency, since you have less I^2 R copper losses in the windings. However, if you are trying to slow to a stop, then applying a meager amount of regen torque won't increase the total amount of regen energy captured because even though the hub motor/controller may be converting a small amount of kinetic energy into battery energy with very high efficiency, you'll meanwhile be loosing much more energy to air drag and rolling friction while you come to a prolonged stop. Likewise, if you come to a very abrupt regen stop, you won't loose much of your kinetic energy to air and rolling friction, but the efficiency of converting that kinetic energy into battery energy via regen will be quite poor. Somewhere in between these two points is some optimum regen braking torque which would get a maximum amount of energy back into the pack.

To get an idea of where it was at I did a series of tests where I rode the ebike up to 40 kph on flat ground, then I engaged the proportional regen in a tucked position and let the ebike decellerate to a stop, this time recording the data at 5 Hz. The results from a typical test are shown below. Here at distance = 0 is where I first engaged the regen. The yellow line is the kinetic energy of me and the bike (1/2 mv^2, where m = 98.7 kg) while the purple line is the cumulative amount of regen energy put in the battery pack. You can see that over the course of coming to a stop from 40kph, I spanned a distance of over 160 meters, starting off with 1.6 Wh of kinetic energy, and ending with about 0.55 Wh of electrical energy back in the battery.

Decelleration with Minimal Regen.gif
Decelleration with Minimal Regen.gif (4.98 KiB) Viewed 8480 times


A test with significantly more regen is shown below, here coming to a stop from 40 kph in just 65 meters:

Decelleration with Significant Regen.gif
Decelleration with Significant Regen.gif (5.04 KiB) Viewed 8479 times


This experiment was repeated a total of 5 times in each case with the regen throttle held steady at a different position. The net results are summarized in the graph below, where I have plotted the percentage of the intial kinetic energy that was returned to the battery as a function of the peak regen current for that particular regen throttle position.
Kinetic Recapture Efficiency.gif
Kinetic Recapture Efficiency.gif (5.7 KiB) Viewed 8469 times


You can see that there is a pretty wide range from about 6 amps to 12 amps of regen current where half of the original kinetic energy was recovered. I was rather surprised that the result was this high and over such a broad range, and 6-12 amps is a good figure since it is within the scope of what most modern rechargeable battery chemistries can take for maximum charge current (~0.5-1C, with ebike packs averaging about 10 Ah).

Justin
Big Dummy with prototype 26" Grin thru-axle front hub, 40A field oriented controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled
User avatar
justin_le
100 kW
100 kW
 
Posts: 1106
Joined: Mon May 28, 2007 3:27 am
Location: Vancouver

Re: Analysis of regen on an ebike

Postby paultrafalgar » Mon Dec 15, 2008 6:02 am

Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
viewtopic.php?f=6&t=7511&start=0&st=0&sk=t&sd=a&hilit=JCG
If we knew what it was we were doing, it would not be called research, would it? Albert Einstein
A paranoid is someone who has SOME idea of what's going on. Allen Ginsberg(?)
If the greatest pleasure is giving, be selfish - give pleasure.
User avatar
paultrafalgar
100 kW
100 kW
 
Posts: 1150
Joined: Fri Feb 15, 2008 5:41 am
Location: Forest of Dean, UK

Re: Analysis of regen on an ebike

Postby justin_le » Mon Dec 15, 2008 6:22 am

paultrafalgar wrote:Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
viewtopic.php?f=6&t=7511&start=0&st=0&sk=t&sd=a&hilit=JCG


Nope, that wouldn't do squat. The energy lost due to the internal resistance of the battery is fairly negligible, and that's the only thing that an ultracap would help with (ie. 'stiffening up' the supply, but most rechargeable batteries are pretty low impedance supplies to begin with) . Of the 50% energy that wasn't recaptured, a majority of that either went to air/rolling drag in the case of using small amounts of regen, or it went into I^2R losses inside the hub motor in the case of high amounts of regen. To get a higher capture of kinetic energy, you would want to have a more streamlined vehicle with higher pressure tires. More mass would help as well.

Justin
Big Dummy with prototype 26" Grin thru-axle front hub, 40A field oriented controller, and 52V 11Ah Cellman triangle pack
My website: http://www.ebikes.ca
Please contact via email, info@ebikes.ca, rather than PMs, which are disabled
User avatar
justin_le
100 kW
100 kW
 
Posts: 1106
Joined: Mon May 28, 2007 3:27 am
Location: Vancouver

Re: Compared to the rolling drag torque

Postby Miles » Mon Dec 15, 2008 6:54 am

justin_le wrote: There are important psychological reasons why freewheeling would be preferred for a lot of people in this case, but the energy argument doesn't seem to hold.


Well, not just psychological... :)

Some people want to be able to ride without using the motor and without any additional drag...

It seems a geared hub-motor with lockable freewheel is needed.....
User avatar
Miles
100 GW
100 GW
 
Posts: 10706
Joined: Fri Mar 16, 2007 4:15 pm
Location: London UK

Re: Analysis of regen on an ebike

Postby snowcrow » Mon Dec 15, 2008 6:58 am

justin_le wrote:
paultrafalgar wrote:Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
http://endless-sphere.com/forums/viewto ... &hilit=JCG


Nope, that wouldn't do squat. The energy lost due to the internal resistance of the battery is fairly negligible, and that's the only thing that an ultracap would help with (ie. 'stiffening up' the supply, but most rechargeable batteries are pretty low impedance supplies to begin with) . Of the 50% energy that wasn't recaptured, a majority of that either went to air/rolling drag in the case of using small amounts of regen, or it went into I^2R losses inside the hub motor in the case of high amounts of regen. To get a higher capture of kinetic energy, you would want to have a more streamlined vehicle with higher pressure tires. More mass would help as well.

Justin


Would a lower system voltage (36v or 48v), be more effective % wise, when using regen?

Blessings, Snow Crow
Dashboard - Sigma BC 506 w/ "Watt's Up" meter
Frame - Schwinn S60 DSX (Re-build in progress)
Wheel Size - 26"
Motor - Crystalyte Rear 408 Hub
Controller - Crystalyte 48v 25amp ver.2
Bike Weight - 80 lbs.
Battery - 48v 12Ah AGM SLA
Total Milage - 140mi
User avatar
snowcrow
100 W
100 W
 
Posts: 167
Joined: Wed Jul 09, 2008 7:27 am
Location: Northeast Kingdom, Vermont

Re: Analysis of regen on an ebike

Postby ZapPat » Mon Dec 15, 2008 8:56 am

Thanks for the really great regen analysis, Justin! I love to see savvy people putting their heads to making good reports like this. I was about to post this over in the first regen thread, but I'll put it here instead since my views of regen coincide with yours.

I can see that there are mixed feelings of the idea of a regen-capable ebike. I guess that people were expecting regen to give them greatly extended range or something even when on flat routes, because the biggest complaint is "it'll only give you an extra 5-10% range". Of course if you are hardly using your brakes in the first place, then regen won't help your range much for sure! I also feel that most problems related to regen are due to poor system design, and as such can be solved by some thought being put into these problems:

- Possible damage to batteries (some suggest it's from overcurrent - no mention of the very possible cell overvoltage problem (overcharge))
A few state (often just repeating others) that using regen automaticly damages batteries. Excessive current is almost always stated here as the culprit, but I believe that over charging the weakest cell is much more of a danger. Of course when planing a regen capable setup we need the choose the right batteries, and set the max regen current appropriatly for what they can take. See my controller suggestion thread where I suggested some ideas for making a foolproof regen setup, consisting of a special controller and BMS working together.

- Motor heating
The efficiency of the motor working as a generator is not much worse than when driving the motor per equivalent watt in/out, so that at 10% average regen this would not represent much additional heat. One real cause of motor heat I've seen reported here on ES is from using a regen controller that doesn't permit freewheeling while regen is activated. This is not a good implementation of regen, and is something that needs improvement.

- Added complexity
Other than bi-directional current sensing and better firmware there's not much to it (see Justin's great explanation). Granted to make it foolproof, some minor modifications to a BMS would also be necessary, but it's not so much added complexity than just making the BMS regen compatible. Once again, see my controller suggestion thread about the BMS ideas.

- Dropout problems
I'm sure this can easily be solved by being aware of certain criteria when choosing a frame for regen. Torque arms are also required for sure, and I would go for two with regen. And maybe some way of locking the nuts too?... Reliable solutions will have to be found for this problem.

- Regained energy ratio only appreciable when used in hills and city driving
I think that regen on light ebikes is most usefull as a brake. Any extra range you get is just a bonus! Some folks will get more bonuses, and some will get less.


On the other hand, the positive aspects are:
- Much reduced wear on brakes
- Smoother braking action (when using current based regen with space-vector modulation)
- Some free extra range gained (proportional to hills and stops in route, along with total weight)

I think Justin has pretty much covered the pros and cons in more detail than I did here, but though I might add my 0.02$ worth anyways.
Pat
All my ebikes have gone up in flames (with my whole house) --> Many lessons learned (like that insurance companies can be cold hearted a$$holes; and to read and understand your contract before your house burns down...)
User avatar
ZapPat
100 kW
100 kW
 
Posts: 1010
Joined: Sat Jun 28, 2008 1:24 pm
Location: Eastern Canada (Gaspésie)

Re: Analysis of regen on an ebike

Postby Miles » Mon Dec 15, 2008 9:05 am

ZapPat wrote:I think that regen on light ebikes is most usefull as a brake. Any extra range you get is just a bonus! Some folks will get more bonuses, and some will get less.

On the other hand, the positive aspects are:
- Much reduced wear on brakes
- Smoother braking action (when using current based regen with space-vector modulation)
- Some free extra range gained (proportional to hills and stops in route, along with total weight)


This sums up why I'm interested in regen, very well.
User avatar
Miles
100 GW
100 GW
 
Posts: 10706
Joined: Fri Mar 16, 2007 4:15 pm
Location: London UK

Re: Analysis of regen on an ebike

Postby paultrafalgar » Mon Dec 15, 2008 9:21 am

justin_le wrote:
paultrafalgar wrote:Justin, could you improve the regen capture by using an ultracapacitor as intermediary, with an ultracapacitor such as JCG uses here:
viewtopic.php?f=6&t=7511&start=0&st=0&sk=t&sd=a&hilit=JCG


Nope, that wouldn't do squat. The energy lost due to the internal resistance of the battery is fairly negligible, and that's the only thing that an ultracap would help with (ie. 'stiffening up' the supply, but most rechargeable batteries are pretty low impedance supplies to begin with) . Of the 50% energy that wasn't recaptured, a majority of that either went to air/rolling drag in the case of using small amounts of regen, or it went into I^2R losses inside the hub motor in the case of high amounts of regen. To get a higher capture of kinetic energy, you would want to have a more streamlined vehicle with higher pressure tires. More mass would help as well.

Justin


ZapPat wrote:- Possible damage to batteries (some suggest it's from overcurrent - no mention of the very possible cell overvoltage problem (overcharge))
A few state (often just repeating others) that using regen automaticly damages batteries. Excessive current is almost always stated here as the culprit, but I believe that over charging the weakest cell is much more of a danger. Of course when planing a regen capable setup we need the choose the right batteries, and set the max regen current appropriatly for what they can take. See my controller suggestion thread where I suggested some ideas for making a foolproof regen setup, consisting of a special controller and BMS working together.


Is there a case for Ultracaps to prevent this overcharging by regen. The ultracap would absorb the energy initially and then transfer it at a non-damaging rate to the battery?
If we knew what it was we were doing, it would not be called research, would it? Albert Einstein
A paranoid is someone who has SOME idea of what's going on. Allen Ginsberg(?)
If the greatest pleasure is giving, be selfish - give pleasure.
User avatar
paultrafalgar
100 kW
100 kW
 
Posts: 1150
Joined: Fri Feb 15, 2008 5:41 am
Location: Forest of Dean, UK

Re: Analysis of regen on an ebike

Postby ZapPat » Mon Dec 15, 2008 9:32 am

snowcrow wrote:Would a lower system voltage (36v or 48v), be more effective % wise, when using regen?
Blessings, Snow Crow

System voltage shoudn't be a particular concern when using regen, other than the usual I^2*R losses. However, regen relies on a PWM boost configuration to work, and this type of circuit gets quite inefficient when the duty cycle gets low. This situation comes up when the motor's back EMF is way lower than the battery's voltage - in other words when we are rolling along at a fraction of our full-throttle speed and want to do regen. The voltage step-up required is then very high, making regen less efficient at these reduced speeds.

So in a nutsheel; choose a system voltage that will let you travel at full throttle comfortably most of the time, because then there is less voltage stepping up to do in the controller for regen to happen. An added bonus to this is that full throttle in most controllers means also less switching losses while driving (if not in current limit mode that is).

Of course even better would be having relays that would dynamicaly switch batteries from series to parallel configurations, and even motor windings from delta to wye for a truly efficient system! This is in the dream real for now, but I just added it to my own controller features wish list! :D
All my ebikes have gone up in flames (with my whole house) --> Many lessons learned (like that insurance companies can be cold hearted a$$holes; and to read and understand your contract before your house burns down...)
User avatar
ZapPat
100 kW
100 kW
 
Posts: 1010
Joined: Sat Jun 28, 2008 1:24 pm
Location: Eastern Canada (Gaspésie)

Re: Analysis of regen on an ebike

Postby Miles » Mon Dec 15, 2008 9:42 am

ZapPat wrote:Of course even better would be having relays that would dynamicaly switch batteries from series to parallel configurations, and even motor windings from delta to wye for a truly efficient system! This is in the dream real for now, but I just added it to my own controller features wish list! :D

Or variable gearing!
User avatar
Miles
100 GW
100 GW
 
Posts: 10706
Joined: Fri Mar 16, 2007 4:15 pm
Location: London UK

Re: Analysis of regen on an ebike

Postby ZapPat » Mon Dec 15, 2008 9:48 am

paultrafalgar wrote:Is there a case for Ultracaps to prevent this overcharging by regen. The ultracap would absorb the energy initially and then transfer it at a non-damaging rate to the battery?

As Justin said, there is no use in using super caps if you have appropriate batteries anyways. The only way you could get around this is having extra circuitry to make a bi-directionnal DC-DC converter between your cap bank and your batteries. I wouldn't consider going this route, specially since real super caps as we need them are still a thing of dreams for the moment, requiring a huge volume to be of any use. Just program your max regen current to something your batteries can handle, and make sure you don't regen when close to a full charge to avoid quickly killing a cell.
All my ebikes have gone up in flames (with my whole house) --> Many lessons learned (like that insurance companies can be cold hearted a$$holes; and to read and understand your contract before your house burns down...)
User avatar
ZapPat
100 kW
100 kW
 
Posts: 1010
Joined: Sat Jun 28, 2008 1:24 pm
Location: Eastern Canada (Gaspésie)

Re: Analysis of regen on an ebike

Postby fechter » Mon Dec 15, 2008 9:57 am

Great job Justin!
That's the best objective analysis I've seen with actual bike data to back it up. I stickified the thread.
There is really not much excuse for not having this feature in all brushless controllers as the additional hardware needed is practically zero. It's just logic. Using synchronous rectification both for regen and drive will improve efficiency and reduce heating as well (another feature commonly missing from most cheap controllers). One downside is the additional complexity may make the controller less "fault tolerant", but this can be addressed with proper design so that processor glitches can't cause shoot-through, etc.
"One test is worth a thousand opinions"
User avatar
fechter
100 GW
100 GW
 
Posts: 10054
Joined: Sun Dec 31, 2006 3:23 pm
Location: California Bay Area, USA

Re: Analysis of regen on an ebike

Postby paultrafalgar » Mon Dec 15, 2008 10:00 am

Miles wrote:
ZapPat wrote:Of course even better would be having relays that would dynamicaly switch batteries from series to parallel configurations, and even motor windings from delta to wye for a truly efficient system! This is in the dream real for now, but I just added it to my own controller features wish list! :D

Or variable gearing!

Hope this doesn't divert the topic, but what's your take on the NuVinci for that, Miles?
If we knew what it was we were doing, it would not be called research, would it? Albert Einstein
A paranoid is someone who has SOME idea of what's going on. Allen Ginsberg(?)
If the greatest pleasure is giving, be selfish - give pleasure.
User avatar
paultrafalgar
100 kW
100 kW
 
Posts: 1150
Joined: Fri Feb 15, 2008 5:41 am
Location: Forest of Dean, UK

Re: Analysis of regen on an ebike

Postby Miles » Mon Dec 15, 2008 10:07 am

paultrafalgar wrote:Hope this doesn't divert the topic, but what's your take on the NuVinci for that, Miles?


I should imagine the efficiency loss in the NuVinci would cancel out the gains in the regen :mrgreen:
User avatar
Miles
100 GW
100 GW
 
Posts: 10706
Joined: Fri Mar 16, 2007 4:15 pm
Location: London UK

Re: Analysis of regen on an ebike

Postby dennis » Tue Dec 16, 2008 8:23 am

Miles wrote:
ZapPat wrote:I think that regen on light ebikes is most usefull as a brake. Any extra range you get is just a bonus! Some folks will get more bonuses, and some will get less.

On the other hand, the positive aspects are:
- Much reduced wear on brakes
- Smoother braking action (when using current based regen with space-vector modulation)
- Some free extra range gained (proportional to hills and stops in route, along with total weight)


This sums up why I'm interested in regen, very well.


All of the above is true, except for the smooth part. With my Bionx system, I found that when braking the whole back wheel vibrates during regen and the back and forth torque placed on the rear axel will cause the nut to loosen over time. I always check my nuts now before each ride :wink: A good solution maybe a custom rear axel for electric hub motors...larger locking nut with a cotter pin with more surface area, similar to an off road motorcycle.... and if you like to track stand http://en.wikipedia.org/wiki/Track_stand, the way the hub sticks when regen kicks in, makes track standing much harder but is possible.

Regen, imo also reduces the useful life of the battery, in my case is LiMn which has a useful life of about 500 full cycle charges. With heavy use, my 9ah battery was only able to hold less than half a charge after 1.5 year. That is why I'm so interested in super capacitors. If someone can put it in a kit without adding too much weight. my 1/100 cent....
2007 Dahon MU SL, Bionx PL500HS, Schlumpf High Speed Drive
User avatar
dennis
10 W
10 W
 
Posts: 83
Joined: Sun Apr 13, 2008 7:00 pm
Location: California

Re: Analysis of regen on an ebike

Postby ZapPat » Tue Dec 16, 2008 10:13 am

dennis wrote:
ZapPat wrote:- Smoother braking action (when using current based regen with space-vector modulation)

All of the above is true, except for the smooth part. With my Bionx system, I found that when braking the whole back wheel vibrates during regen and the back and forth torque placed on the rear axel will cause the nut to loosen over time. I always check my nuts now before each ride :wink: A good solution maybe a custom rear axel for electric hub motors...larger locking nut with a cotter pin with more surface area, similar to an off road motorcycle.... and if you like to track stand http://en.wikipedia.org/wiki/Track_stand, the way the hub sticks when regen kicks in, makes track standing much harder but is possible.

I would agree with you on this point for regular regen, but as you see I did specify using SVM which reduces torque ripple both during drive and regen. This happens because a SVM capable controller will output a waveform that matches the motor's own back EMF, thus avoiding the uneven current flows that regular trapezoidal commutation produces. This of course does require more complexity mostly in firmware, more processing power, faster ADCs and phase current sensing. I must admit I have no idea what biox uses as a regen technique, but from what you say they must be using regular old commutation techniques.

Of course even with SVM we still have the bidirectionnal torque issue (although probably lessened with the smoother SVM) caused by regen. This will just be a matter of having good torque arms, along with some setup that will effectively lock your nuts to prevent them from loosening. I'm not a hardcore bike guy, so I'll leave this part for other great brains here to solve. Note that Justin just tighted the hell out of his nuts and that was already enough for his setup to not come loose, so I don't think that the solutions will be that complex.

dennis wrote:Regen, imo also reduces the useful life of the battery, in my case is LiMn which has a useful life of about 500 full cycle charges. With heavy use, my 9ah battery was only able to hold less than half a charge after 1.5 year. That is why I'm so interested in super capacitors. If someone can put it in a kit without adding too much weight. my 1/100 cent....

Read Justin's posts, along with mine, then explain why you claim that regen inherantly reduces battery life when current limits and such are well thought out? In other words, explain what is the difference between regenerated current and regular charging current going into the battery? Also, what makes you sure that something else was not the cause of your packs premature demise? I know I might sound rash a bit here, but I strongly dislike unsubstantiated claims being posted by a few people and then repeated by many others... :evil:

Cheers!
Pat
Last edited by ZapPat on Tue Dec 16, 2008 3:30 pm, edited 1 time in total.
All my ebikes have gone up in flames (with my whole house) --> Many lessons learned (like that insurance companies can be cold hearted a$$holes; and to read and understand your contract before your house burns down...)
User avatar
ZapPat
100 kW
100 kW
 
Posts: 1010
Joined: Sat Jun 28, 2008 1:24 pm
Location: Eastern Canada (Gaspésie)

Re: Analysis of regen on an ebike

Postby fechter » Tue Dec 16, 2008 10:28 am

ZapPat wrote:
dennis wrote:Regen, imo also reduces the useful life of the battery, in my case is LiMn which has a useful life of about 500 full cycle charges. With heavy use, my 9ah battery was only able to hold less than half a charge after 1.5 year. That is why I'm so interested in super capacitors. If someone can put it in a kit without adding too much weight. my 1/100 cent....



I agree. Unless you have some kind of side-by-side test with careful controls, it would be hard to blame short cycle life on regen. My hybrid car has Nimh batteries and uses regen extensively. It has 89,000mi on it now and many thousands of cycles. If the regen current was too high and allowed the cells to get hot, then that would be a reason you could blame regen for shortened life.
No matter what the chemistry, the regen circuit must keep battery currents and voltages within the safe range for the batteries used.
"One test is worth a thousand opinions"
User avatar
fechter
100 GW
100 GW
 
Posts: 10054
Joined: Sun Dec 31, 2006 3:23 pm
Location: California Bay Area, USA

Re: Analysis of regen on an ebike

Postby MitchJi » Wed Dec 17, 2008 1:08 am

Hi,

dennis wrote:Regen, imo also reduces the useful life of the battery, in my case is LiMn which has a useful life of about 500 full cycle charges. With heavy use, my 9ah battery was only able to hold less than half a charge after 1.5 year. That is why I'm so interested in super capacitors. If someone can put it in a kit without adding too much weight. my 1/100 cent....


ZapPat wrote:Read Justin's posts, along with mine, then explain why you claim that regen inherantly reduces battery life when current limits and such are well thought out? In other words, explain what is the difference between regenerated current and regular charging current going into the battery? Also, what makes you sure that something else was not the cause of your packs premature demise? I know I might sound rash a bit here, but I strongly dislike unsubstantiated claims being posted by a few people and then repeated by many others... :evil:


I think Dennis is concerned that one metric of cell life is charge/discharge cycles and with regen there are a lot more (small) charges. In other words I don't think his comment has anything to do with any "difference between regenerated current and regular charging current going into the battery".

I also think "unsubstantiated claims" might be a bit strong. He didn't actually state that his pack life was reduced by regen. He said with heavy use his pack only lasted 1.5 years and I think he is concerned that regen might reduce his pack life even further.
Best Wishes!

Mitch


This table compares the cost/wh and lifetime cost based on a cycle life of 500 cycles. Nissan Leaf Cells beat Ping LiFePO4 and HK LiPo hands down!

Leaf Cells, tier 1 OEM quality Lithium for roughly a similar lifetime cost as lead!

Module specs are here

Get a great deal on an EV (NEV), buy my ZENN!
"2006 ZENN NEV - Only ~2K Miles - Excellent Cond! - Only $2k"
User avatar
MitchJi
1 GW
1 GW
 
Posts: 3037
Joined: Mon Jun 02, 2008 8:09 pm
Location: Marin County California

Re: Analysis of regen on an ebike

Postby solarbbq2003 » Wed Dec 17, 2008 7:51 am

nice one justin!
another interesting thing is you can capture more regen when the batteries are more discharged, its quite noticeable, I guess the batts have lower resistance the more discharged they get and hence can capture more regen current, not sure if there is any practical way to utilise that, apart from having a spare pack which is discharged and put the regen back into that pack,
solarbbq2003
10 kW
10 kW
 
Posts: 535
Joined: Sat Apr 07, 2007 10:21 am

Re: Analysis of regen on an ebike

Postby solarbbq2003 » Wed Dec 17, 2008 7:53 am

some advantages though for geared hub motors with no freewheel as can produce alot more amps in regen than a non geared
solarbbq2003
10 kW
10 kW
 
Posts: 535
Joined: Sat Apr 07, 2007 10:21 am

Re: Analysis of regen on an ebike

Postby ZapPat » Wed Dec 17, 2008 9:17 am

MitchJi wrote:Hi,
dennis wrote:Regen, imo also reduces the useful life of the battery, in my case is LiMn which has a useful life of about 500 full cycle charges. With heavy use, my 9ah battery was only able to hold less than half a charge after 1.5 year. That is why I'm so interested in super capacitors. If someone can put it in a kit without adding too much weight. my 1/100 cent....

ZapPat wrote:Read Justin's posts, along with mine, then explain why you claim that regen inherantly reduces battery life when current limits and such are well thought out? In other words, explain what is the difference between regenerated current and regular charging current going into the battery? Also, what makes you sure that something else was not the cause of your packs premature demise? I know I might sound rash a bit here, but I strongly dislike unsubstantiated claims being posted by a few people and then repeated by many others... :evil:

I think Dennis is concerned that one metric of cell life is charge/discharge cycles and with regen there are a lot more (small) charges. In other words I don't think his comment has anything to do with any "difference between regenerated current and regular charging current going into the battery".

Hummm... There is no "cycle" counting mecanism in batteries, so smaller charge/discharges are not worse on them than fewer deeper cycles. In fact there seems to be evidence of the opposite, if all other conditions are kept equal. Read Sandia's lifebatt test report and you will see this is true. Fechter also seems to be of this opinion in the previous post (although his post is a bit confusing since he almost seems to be agreeing with denis until you really read what he says - you might want to clear that up, Fechter...).

I also think "unsubstantiated claims" might be a bit strong. He didn't actually state that his pack life was reduced by regen. He said with heavy use his pack only lasted 1.5 years and I think he is concerned that regen might reduce his pack life even further.

Well I might be misreading him, but what does "Regen, imo also reduces the useful life of the battery" sound like to you? Also, Denis talks about super caps helping his battery life when doing regen, which pretty much says he thinks that without them the pack's life is reduced. As discussed in earlier posts on this thread, you would need a fairly complex super-cap setup using an extra bi-directionnal DC-DC converter between the battery and the capacitor bank to have the capacitors help you at all. Just putting a parallel cap bank won't help you, since battery voltage goes up quite slowly while charging, but caps don't have that voltage plateau. So, as Justin says, caps in parallel will just smooth the current a bit, but will not be able to store any significant amount of energy since their voltage can't rise over the batterie's voltage plateau.

At any rate, Denis can respond to defend his claim. All I can say is that the numbers I've seen don't support the claims of reduced cell life being linked directly to regen. A badly done regen setup would shorten cell life for sure, and I think that this is the real problem with almost all present ebike regen systems (note that hybrid cars already have full regen protection, and for good reasons too!)

What is missing almost always in regen setups? Most people using good regen controllers set a max regen current limit, so that will protect the battery from heat-related charging problems caused by too much charge current. This leaves over-voltage protection (over charge protection), but this protection right now is only done on a pack level in the controller, which as we well know can be quickly fatal to the first cell reaching it's end of charge voltage. This is why we have BMS circuits used while charging that offer cell-level protection for charge termination. As far as I know, most regen setups don't go though the BMS, and as such this very, very important protection feature is not present during regen, but only during regular charging. This I think is the real culprit behing the present widely-repeated belief that regen inherantly reduces cell life.

And now what's the solution to the cell over charge problem? I'm looking into having some basic link between the BMS and the regen-capable controller, so that the controller knows as soon as the first cell hits full charge and disables regen in a safe way for the user. What most people rely on right now to avoid this destructive situation is to not use regen when their battery might be close to full charge - but only one error, only once, and blam there goes a cell!

Does this help explain my slight annoyance when reading such claims about regen reducing battery life, but with no explanations to prove the point? If anyone can prove the opposite, please post numbers and I will gladly learn from them! I know I can find numbers to prove my points here... So before dissing something (such as regen here), please understand it first!

Cheers!
Pat
All my ebikes have gone up in flames (with my whole house) --> Many lessons learned (like that insurance companies can be cold hearted a$$holes; and to read and understand your contract before your house burns down...)
User avatar
ZapPat
100 kW
100 kW
 
Posts: 1010
Joined: Sat Jun 28, 2008 1:24 pm
Location: Eastern Canada (Gaspésie)

Next

Return to E-Bike Technical

Who is online

Users browsing this forum: Samd, slawsonxo and 12 guests