Supercapacitor Regen Bank ?

qwerkus

10 kW
Joined
Jul 22, 2017
Messages
785
Hello,

There are numerous studies showing the usefulness of supercapacitor banks to recover brake power in automotive transport. The idea is pretty simple: brake power come in short bursts which cannot be stored in batteries, but supercaps can. Conversly, the startup of any e-motor requires a peak torque: here the stored energy from supercaps comes very handy. Basically an electrical flywheel.

What I'm wondering is if the same principle could be applied to ebikes. Most of the regen energy is lost / cannot be used, especially in stop-and-go situations like city commuting. If we could store some of the energy lost at each stop, and use it again to kick start the motor when the light turns green, we could certainly save some battery power. Since medium capacity Double Layer supercaps are becoming quite cheap, it wont cost an arm and a leg to try it.

A 48V Bank (actually 54) would require 20 supercaps in series. If we use samwha 100F caps, that's around 500g (including balancing circuit), 0.4L and 60$ for 5F total capacity. 200F/cap would double the total capacity to 10F, but I've not found cheap ones yet.

Assuming 5F are loaded to 54V, that's (W = 1/2 C V2 ) 7290J storage capacity.

Assuming the bank is fully loaded, and discharges in 1s, the useful power would be around (P = dW / dt ) 7KW - plenty enough to start a motor.

What do you think ?
 
For some existing info, theres a number of threads discussing the basic idea, such as this one

https://endless-sphere.com/forums/viewtopic.php?f=14&t=91799

finding the others may require a bit of poking around, mostly they shoudl be findable by a search on titles or first posts of threads, displayed by topic, search terms of supercap* or ultracap*, possibly in combination with regen* (but not always).
 
amberwolf said:
For some existing info, theres a number of threads discussing the basic idea, such as this one

https://endless-sphere.com/forums/viewtopic.php?f=14&t=91799

finding the others may require a bit of poking around, mostly they shoudl be findable by a search on titles or first posts of threads, displayed by topic, search terms of supercap* or ultracap*, possibly in combination with regen* (but not always).

I actually did some research before posting, and found a lot of threads mentioning supercaps indeed. From what I gathered, the consensus is that supercaps are no match for li-ion in terms of energy storage. Yet the idea to limit their application to capture break power and give a boost during startups keeps resurfacing as viable. hence me question. Has there been any field test ?
 
The place where adding supercapacitors to a battery would help the most (IMHO) is a DD hubmotor ebike on fairly flat land. On a long downhill, the SC's would be "filled" in a short amount of time, and then the rest of the downhill, any further regen harvesting would be limited by the batteries ability to absorb it, which is the original bottle-neck in the first place.

There are several videos on youtube showing someone who used an SC bank to start a car (with the lead-acid battery disconnected to prove the SC bank is as capable as possible). I finally found one video where they also disconnected the spark and the fuel relay, so the car engine would turn over without actually starting, and the SC bank lasted about 30 seconds. Another video measured the start-up amps, and a typical 4-cylinder with an average compression ratio hit 200A for a half-second...then it continued to draw about 100A continuously.

I'm told that when you double the SC's in in series to increase the volts, it halves the capacitance in farads (range?). I need to learn more about how this would work, but...if a 12V SC bank can crank for 30-seconds, would a 48V SC bank crank for 8 seconds?

A long uphill with a commensurate long downhill on the other side would have a high amp-draw all the way up, and a lot of time on the downhill where regen "could be" harvested. That might take a huge SC bank, and for cost and size considerations, it might be better to just buy a high-amp battery pack.

I don't really know, but...these are just some thoughts...
 
qwerkus said:
Assuming 5F are loaded to 54V, that's (W = 1/2 C V2 ) 7290J storage capacity.
Assuming the bank is fully loaded, and discharges in 1s, the useful power would be around (P = dW / dt ) 7KW - plenty enough to start a motor.

That assumes the cap bank can be drained to 0V to extract all the stored energy, which isn't the case without a powerful DC-DC converter. So you need to recalculate for the voltage range actually used.

I'm not sure whether you intended the caps to the only storage on the bike, or used in parallel with a normal battery?

Either way, most/all ebike batteries cope with all the regen current available on a typical ebike. Bicycles do have quite a limited amount of energy that can be recovered compared to other vehicles.

Also, the weight, volume and cost of supercaps is usually inferior to an equivalent li-ion battery, that happens to store an order of magnitude more energy.
 
Agreed: my math is far from exact, and there will be a lot more number crunching required. But again: this is not a project aimed at wildly increasing range of ebikes with supercaps, but a very limited application of cheap supercapacitor storage in a stop-and-go configuration during commute drive in densely populated area. The main goal is to harvest otherwise lost energy, nothing more, nothing less.

I'm working with a chinese manufacturer to enable regen on a bpm type hub motor: small bldc motor + 1:4 planetary gear. This won't be great for long downhills, as there is no way to properly cool the hub, but it could work in a stop-and-go config. First step will be to measure the amount of power actually produced during braking; if Punx0r is correct, and the li-ion battery can handle the burst, there will be no need for a bank. Though the manufacturer claims otherwise: they allegedly measured a 15A @60V burst during sharp breaking test in lab conditions. That's 900W over 1.5s - more than any bms charger can take.
 
qwerkus said:
There are numerous studies showing the usefulness of supercapacitor banks to recover brake power in automotive transport. The idea is pretty simple: brake power come in short bursts which cannot be stored in batteries, but supercaps can.
In general, I think resizing batteries and/or choosing the right battery specs are an easier way to accomplish this. The electronics to selectively route energy to supercaps rather than batteries is fairly complex, and supercaps have their own issues (i.e. voltage range.)
 
I like this idea. Right now I limit my regen battery current to 20A which is 1C for the multistar battery pack I have. I would definitely like the option to have stronger regen without stressing my battery any more.

Maybe a second ESC could be used for braking only, to direct current into the supercap bank. If you did this you would need to make sure only one ESC is using the motor at a time.
 
Adding some high discharge/charge rate Li battery is "gram per gram, Wh per Wh and cent per cent" much better solution then SC. There are plenty of option in any flavor LiPo, LiFePo, LTO etc, rated 5,10 and more C for charging.
Charge rating is actually for continuous charge, which creates electron overflow and heat, raising voltage and starting irreversible chemical reaction on electrolyte. I have some old Turnigy Hardcases, that every day see 2C regen bursts for last 7 years and with 60% initial capacity, degradation is more calender and cycle life related. Regen charge bursts are only annoying on fully charged battery, where somewhat intelligent controller just stops regen to not over volt the battery.
 
A modest battery can absorb all the regen energy from a scooter you can throw at it.

Running a cap would just mean more loss, cost, and added failure modes.
 
I had a 18fet on a muxus 3,000. My controller was throwing 2,250 watts regen measured on my CA to my 24s A123 20 ah no bms. I couldn't reprogram it and ran it this way for a year. Only a few battery wouldn't mind. Quality matters.
 
Without sophisticated high power control electronics, the supercap bank would only store recoverable energy in a narrow voltage range corresponding to the battery's operating range-- but it would store useless but still hazardous energy from that voltage down to 0V, so most of its capacity. It doesn't seem like a very good deal to me.

The idea might have some merit, but not with supercaps. I'm thinking something like a small capacity but high power LTO pack with slightly higher operating voltage and diode isolation from the main pack, so that regen current goes to to the LTO pack preferentially and then discharges from it preferentially until that smaller pack drops to the same voltage as the main pack (when they begin to work in parallel).

Even if that scheme worked, it would incur losses from the Vf drop in each diode that carries power. So it would have to work very well to more than offset the losses.
 
Thank all for your valuable input. Are there any controller out there with integrated and programmable bms ? It seems to me that to make this thing working, either with supercaps or some other high charge high drain battery I will have to rework the charging circuit of the li-ion pack. There is a lot of research beeing done on fast charging, and one idea would be to tweak the bms, so it would accept 2C up to 70% of battery capacity, and than cut down to 0.5-0.8C (see this article With a 4p or a 5p pack, there should be enough capacity to absorb breaking regen.
 
Regen charges back through the same current path it discharged through to power the vehicle.
 
...so you are not limited by the current rating of the "charging" side of the BMS (5A, 10A etc) :)
 
Punx0r said:
...so you are not limited by the current rating of the "charging" side of the BMS (5A, 10A etc) :)
No you wont be limited by the current rating of the charge input, as the regen current flows through the same wires, only in reverse direction.

There are BMS where both, charge and discharge can be attached to "C-", and other BMS which have sparate "P-" and "C-" for output and charger.
Regarding latter type of BMS it isn't a problem to push regen current also through "P-", BUT there will not be overcharge protection as the "P-" mosfets cannot turn off the current due to the diodes.
So a charger should be always connected to "C-" for safety.
 
liveforphysics said:
Regen charges back through the same current path it discharged through to power the vehicle.

So how does balancing work in regen mode ? Not at all ?
Also it doesn't change the fact that i'd need a programmable controller - or an additional device between the controller and the bms - to regulate regen bursts.
I'm going to have a closer look at the open source VESC. There is a post where user Addy says he successfully implemented progressive regen on a VESC. Maybe that config could be tweaked into a "fast-charge" regulator.
 
qwerkus said:
liveforphysics said:
Regen charges back through the same current path it discharged through to power the vehicle.

So how does balancing work in regen mode ? Not at all ?
Also it doesn't change the fact that i'd need a programmable controller - or an additional device between the controller and the bms - to regulate regen bursts.
I'm going to have a closer look at the open source VESC. There is a post where user Addy says he successfully implemented progressive regen on a VESC. Maybe that config could be tweaked into a "fast-charge" regulator.
Regen pushes back about 2% in my average ride and I would not be solving an in-existent problem of balancing cells at regen (which occurs anyway, because BMS does not care where the current comes from, is it charger or controller). You must be living on peak pike to care where to put excess regen energy in. I live on the hill already and just have mediocre regen the first sloop, because my cheap Xiechang controller has regen limit voltage set, then I hit throttle for 30 meters and next sloop of 200m is not enough to limit my regen by overvolting to set point.

Are you looking for a solution of some real world existent issue or just expanding a theory of hypothetically possible occurrence?
 
parabellum said:
Are you looking for a solution of some real world existent issue or just expanding a theory of hypothetically possible occurrence?

See earlier post: working on a gear hub with regen. Wondering if it is possible to recover the energy from short bursts in a stop and go config. The manufacturer claims power levels are too high for the battery to handle it. Hence this investigation.

I've been doing some maths: if I drive 8.5m/s (30.6km/h) with a combined bike + driver weight of 100Kg, that's roughly: 3612.5 J of kinetic energy. (KE=1/2*m*v^2) If the generator (motor in regen mode) and the loading circuit has an (optimistic) efficiency of 50%, that's 1806J to cope with over 1s (averaged brake time) this means a burst of 1.8KW. So if I'm not mistaken, at least in theory, the manufacturer is correct. But it's true that those numbers are far from experimental results, so I will have to measure the stuff myself before buying new batteries.
 
I get 8% regen average over four years and live with many Hills and I have the the regen torqued up. How much region voltage do you think you're going to get back ? It's biggest Plus is braking with Rim brakes even with my Kool stops.
 
I changed my braking battery current limit to 30A before riding home yesterday. Here's an example from my log of that ride:
acceleration  + brake example.png

That's with 100A phase current limits, 60/30A battery discharge / charge limits on my VESC 4.12. It's hard to tell on the graph but the regen only reached 25A and then quickly declined from there. On this ride I got 5% back from regen. It's a flat route with several stops for traffic.

I also don't have a BMS on this pack, I only charge to 4.1V/cell and I've never fully discharged. I check the balance usually once a month and the cells are all still very close to each other.
 
qwerkus said:
The manufacturer claims power levels are too high for the battery to handle it.

I've been doing some maths: if I drive 8.5m/s (30.6km/h) with a combined bike + driver weight of 100Kg, that's roughly: 3612.5 J of kinetic energy. (KE=1/2*m*v^2) If the generator (motor in regen mode) and the loading circuit has an (optimistic) efficiency of 50%, that's 1806J to cope with over 1s (averaged brake time) this means a burst of 1.8KW

In a literal analysis I think the manufacturer is correct. The average ebike pack is often low-powered, maybe only 3 or 4 parallel low-discharge rate 3400 or 3500mAh cells with a maximum charge rate of 1C (~3.5A) on the datasheet, so any more than 10-12A of regen (entirely plausible) looks like too much.

As pointed out above, though, that 1C is for a continuous charging condition, which is limited by heating of the cell. Burst charge, like burst-discharge, rates are allowed to be greater.

On your maths, I don't think you're too far off in your estimates, but I think your 1.8kW is too high. Braking from 30kph/20mph to zero in 1 second is unrealistic: that's 1g of deceleration and is roughly what a modern car with good tyres and brakes is capable of with good road conditions under emergency braking. You just can't brake that hard on a bicycle without going over the handlebars. A deceleration time of 4-5 seconds is probably more realistic. There's a bit of fudging there, as regen becomes less effective as speed reduces and generally won't bring you to a stop in a reasonable time.

Anyways, even with 1.8kW on an average 12-series (44V) li-ion pack is 41A. If the pack is 4-parallel cells, that's 10A per cell, or ~3C, which is probably fine in short bursts.
 
Slowing aggressively (on the verge of locking rear tire) from about 50mph I can achieve about 60a back into my pack. This naturally only lasts for a few seconds. Decelerating at 4kw is somewhat similar to accelerating with 6kw, given losses etc. I've been doing this daily on this pack for 2+ years, this is substantially higher than the maximum rated charge on Samsung 30Q cells, but as Luke touched on - they don't much care about big regen bursts at moderate SOC.

Quite simply you're solving a problem that doesn't exist on lightweight vehicles.
 
qwerkus said:
The manufacturer claims power levels are too high for the battery to handle it.
My impression is that manufacturers try to reduce warranty cases as much possible and save on every end.
For example, the battery manufacturer (assembler) use separate charge and discharge port to save on charge side fets, cancelling regen responsibility right of the box. If using same output, capable for charge and discharge (regen), fets need to be same specs on both sides and add few cents on end price.
Regen issue is easily solvable and most batteries will tolerate that.
Caps, at current stage, where beaten to death already in multiple treads, I considered caps for inductive load start spike via inverter(seconds mater), with low C rate bats as main source and ended with LTO to take that spike for economical reasons, A123LiFePo was a second choice.
 
Back
Top