Definitive Tests on the Heating and Cooling of Hub Motors

[gensem]

Good point about a mag rim heat sink. If only we could get some dedicated mag rim hub motors for e-bicycles that look cool and are strong and light (will not make it look like a scooter). Maybe real magnesium.

A small and thin cast wheel ll probably add 1-1,5kg vs regular rim + spokes. Depending on your power needs and base hubmotor weight thats feasible.

 

[macribs]

I hope Vito reads this thread, maybe QS can make good looking light weight and strong aluminum or magnesium wheels. With option to choose 17"-19".
I saw in the QS thread there is a 17" alloy wheel with hub motor but it kind of looks too much like a scooter wheel at least for my taste.


I wonder if the RC motors like Revolt 160 Pro would be a good candidate for Ferro Fluid? Either as a mid drive or a swing arm mounted motor to get the weight out of the hub? Custom made side covers and FF?? - I think in RC air plane they expect 15-20kw peak from the new Revolt 160. Wonder what kind of peak performance we could see when cooled with FF? Say WOT from standstill for 5-10 seconds? Repeatedly over and over possible or will this motor not be a good candidate for FF?

 

[fechter]

Are those motors outrunners? I'm not sure FF will behave well if the motor RPM is too high. I might try some crude testing with an old brushed motor. I have some FF that is made with tiny particles of NdFeB. This stuff won't saturate easy.

Justin; just curious about the potting. Did you weigh the stator before and after to see how much weight the potting adds?
That would be the main downside of potting. Having the halls mounted on a curved piece of circuit board that screws into the stator is a handy way to make hall replacement easier. The old Kollmorgen motors had something like this.

On hall failures, I think the primary fail mode after cable damage (spin out) is when the 5v line contacts the signal line and the output is trying to go low. This causes the current to exceed the ratings and smoke it. If the 5v hall supply line had a current limiter that was set below the rating, the halls wouldn't be able to fry by any kind of misconnection. This limiter could be anywhere between the controller and the hall sensors. Placing it in the controller might make more sense but it could also be in the motor. A simple resistor in series with the 5v line might work.

 

[speedmd]

Are those motors outrunners? I'm not sure FF will behave well if the motor RPM is too high.

The revolt 160 is a open sided out runner. Looks to have very good air flow potential. Don't think you would want to add anything that could catch and hold dust.

 

[justin_le]

Are those motors outrunners? I'm not sure FF will behave well if the motor RPM is too high.

The revolt 160 is a open sided out runner. Looks too have very good air flow potential. Don't think you would want to add anything that could catch and hold dust.

Agreed completely. A really high RPM open faced motor with good airflow is not really the kind of candidate that jumps out for needing or wanting FF. If you have to seal the motor for environmental exposure reasons, then there would be much more of a case for it. I will be looking into this application in R/C style outrunners in the context of skateboard hub motors where it's really best to seal things, but overheating then becomes a big concern (especially as there is so little thermal mass to absorb heat from short steep hill climbs). So it's on the testing agenda.

Justin; just curious about the potting. Did you weigh the stator before and after to see how much weight the potting adds?

Nope, but I weighed how much material I mixed up and poured in and that was around 300 grams, vs a total motor weight of 9kg. So we're looking at a 3% mass increase for the hub.

That would be the main downside of potting. Having the halls mounted on a curved piece of circuit board that screws into the stator is a handy way to make hall replacement easier. The old Kollmorgen motors had something like this.

Yup for sure, that's a possible approach. But going forwards halls are going to be retired from hub motors just like we retired brushes from hub motors about 10 years ago. Unless you're making a self balancing style vehicle that needs accurate torque control at 0rpm, there's just no need for hall sensors with a modern controller. So I'm not going to go too far out of my way in trying to accommodate halls or hall replacement strategies if it turns out that potting is indeed beneficial enough to justify the extra steps and expense.

On hall failures, I think the primary fail mode after cable damage (spin out) is when the 5v line contacts the signal line and the output is trying to go low. This causes the current to exceed the ratings and smoke it. If the 5v hall supply line had a current limiter that was set below the rating, the halls wouldn't be able to fry by any kind of misconnection. This limiter could be anywhere between the controller and the hall sensors. Placing it in the controller might make more sense but it could also be in the motor. A simple resistor in series with the 5v line might work.

Yes exactly, it doesn't take too much to protect a signal line against all the possible voltages it could see in the worst kind of axle spinout mishap or faulty connection pinout. The old analog crystalyte controllers had the halls power from a 12V bus with an inline resistor, such that once the 3 halls were connected then the resistance + quiescent current dropped the voltage on hall Vcc to ~6V. In those units a worst case current if the signal is shorted to the halls' Vcc was only ~20-30mA. Anyways this is all to say I'm not super concerned about the need for making halls easily replaceable at this stage.

 

[themadhatter106]

It's great that your doing this.

Once we have a better thermal conductivity to the outside shell what then? Will you test increasing the surface area of the rotor in some way? Maybe attaching heat sinks?

 

[John in CR]

A hubmotor with integrated aluminium rim should transfer alot more heat with FF

One thing that must be taken into account about heat transfer is that with no exterior changes (shell, environment temp, air flow, etc), the only way to move more heat is that the outside surfaces must be hotter. That means everything in the heat pathway all the way back to the copper and stator steel must be hotter too. Yes, Justin's thorough tests of FF's effects (thank you Justin, because I initially dismissed how much effect FF could have on the heat pathway.) clearly showed a cooler stator for the 160W or so of heat dissipated to the environment. Increase the amount of heat and the temperature of everything goes up without exterior changes.

Take Itchynackers Pike's Peak climb for example. He ran a healthy fill of ATF in the hubbie, so he had a good thermal connection between the stator and inside of the motor shell, similar to FF. In terms of heat, the drag of the oil did add a minor amount of heat to the equation that Justin has shown FF adds only a tiny drag. Mr. Nackers made no outer shell changes, and ran the climb at about 3kw input if memory serves correctly. Also of note is that the temperature was around freezing, and he added a healthy amount of human assist to the climb. Despite the cold temps, and what is considered by so many to be better cooling, he still had heat issues with stator temps pushing past 130°C. Sure the stator would have been hotter with no oil, but it obviously still wasn't nearly effective enough, and that was at nowhere near the torture of typical off road riding with uphill grades. Relatively easy changes to the outside to both add surface area and increase turbulent flow at those surfaces working in conjunction with the oil fill would have resulted in greatly reduced stator temps and greater efficiency (less heat to dissipate)...all for an insignificant increase in aerodynamic drag.

Macribs,

FWIW, increased thermal mass at the covers that you mentioned offers only a short term improvement early in each ride. You only want to add whatever thermal mass is necessary to achieve the increase in surface area that is effective. ie You don't want the covers absorbing and holding more heat. Instead you want them to dissipate heat faster at lower temperatures, so the interior surfaces are cooler and can therefore accept heat faster. Also, keep in mind that heat transfer is ultimately dependent upon the temperature differential between the environment and outer shell temp, so when looking at ICE solutions keep in mind our temps are drastically lower. That's why liquid cooling solutions involving cooling fluids flowing through interior manifolds and pumped outside to radiators never work as well as most people think, and there have been some pretty good attempts.

 

[justin_le]

Wonder what kind of peak performance we could see when cooled with FF? Say WOT from standstill for 5-10 seconds?

In case people are being a bit misled, the so called "peak performance" of a motor won't really change AT ALL from having ferrofluid or any other cooling strategy employed. Please get this straight, all that would be improved is the continuous sustained power handling capability of the hub over long durations. When people talk about peaks of 10's of kW or whatever, that is almost 100% being taken care of by the thermal mass / heat capacity of the motor. For bursts of 5-10 seconds the conductivity is totally moot. What would be improved is the duty cycle with which you could repeat those massive power bursts without eventually overheating since things would cool faster after each burst, but the peak itself? No change.

It's kindof like comparing the anaerobic vs. aerobic muscle output. In a fury of adrenaline you can easily put out over 700-800 watts of human power until lactic acid buildup is too much, regardless of how much oxygen is getting to your bloodstream. But over longer time, the your muscle power output is really going to be limited by the amount of oxygen uptake you absorb in your system. Adding ferrofluid/atf/vent holes etc. is like increasing your maximum oxygen uptake, rather than increasing your muscle mass. You'll be able to run harder and for longer, but your sprint won't improve at all.

 

[macribs]

Are those motors outrunners? I'm not sure FF will behave well if the motor RPM is too high. I might try some crude testing with an old brushed motor. I have some FF that is made with tiny particles of NdFeB. This stuff won't saturate easy.

Yes those Revolt motors are outrunners. KV of 45. So with a 20s pack about ten times the speed of a hubmotor. FF might be flung everywhere when motor RPM increased?

The revolt 160 is a open sided out runner. Looks to have very good air flow potential. Don't think you would want to add anything that could catch and hold dust.

Yes it seems airflow should be good, however I wonder how that motor would work if closed with side covers, and filled with FF. Reason for FF and closed side covers would be to keep salty humid air away for motor internal, as well as avoid splashing of saltwater during winter season when they cover all roads with sea salt to avoid icy conditions.

I guess what is my concern is if the Ferro Fluid would behave different if used with an much faster spinning RC outrunner then FF does behave in way slower turning hubmotors.

A really high RPM open faced motor with good airflow is not really the kind of candidate that jumps out for needing or wanting FF. If you have to seal the motor for environmental exposure reasons, then there would be much more of a case for it.

That is just the case I am thinking of. Living on the coastline we have a salty environment, with high amount of saltwater in the air year around. If you look at cars driven here where I live they are more likely to develop corrosion problems then cars driven inland. Then add the fact that the rads get covered in salt during the long winter making the snow and ice on the road nasty, salty slush and water the can penetrate almost anything.


I guess that was exactly what I was trying to point out, hence the wording repeatedly over and over. I did choose me words poorly I can see that.
What I was really asking was if the FF would aid in the cooling of modified RC outrunners when running WOT for 5-10 seconds repeatedly. From another thread it was mention that people should not expect to reach same burst of power from RC motors when used in ebikes as RC motor(s) would have way less cooling on a bike versus an airplane.

 

[spinningmagnets]

To compare the two (large hub motor vs small RC outrunner), the outrunner has a thin steel shell, that factor coupled with its smaller diameter means it has much less heat-absorbing mass, and much less surface area to shed any heat it does absorb. Given these characteristics, I agree that an RC outrunner would not benefit much from FF.

I'm sure there would be some measurable benefit, but it seems likely that the benefit would be small. The good news is that, adding a temp sensor and a few cc's of FF is fairly affordable and easy for anyone who wants to pursue gathering that data.

 

[macribs]

To compare the two (large hub motor vs small RC outrunner), the outrunner has a thin steel shell, that factor coupled with its smaller diameter means it has much less heat-absorbing mass, and much less surface area to shed any heat it does absorb. Given these characteristics, I agree that an RC outrunner would not benefit much from FF.

I'm sure there would be some measurable benefit, but it seems likely that the benefit would be small. The good news is that, adding a temp sensor and a few cc's of FF is fairly affordable and easy for anyone who wants to pursue gathering that data.

Take a look at the pic below, AFAIK this motor is one of the largest RC outrunners I have seen so far. The size of the motor made me think it could be a candidate for FF.

 

[Offroader]

For the FF to work does the motor need to be spinning? I'm not sure if where the FF pools if it will make contact with the stator when the motor is not spinning.

 

[bowlofsalad]

For the FF to work does the motor need to be spinning? I'm not sure if where the FF pools if it will make contact with the stator when the motor is not spinning.

https://endless-sphere.com/forums/download/file.php?id=178687

Ferrofluid is a fluid that is influenced my magnets, it sticks to magnets (better phrasing, magnets hold onto this fluid), very different than just a plain old liquid.

It's kindof like comparing the anaerobic vs. aerobic muscle output. In a fury of adrenaline you can easily put out over 700-800 watts of human power until lactic acid buildup is too much, regardless of how much oxygen is getting to your bloodstream. But over longer time, the your muscle power output is really going to be limited by the amount of oxygen uptake you absorb in your system. Adding ferrofluid/atf/vent holes etc. is like increasing your maximum oxygen uptake, rather than increasing your muscle mass. You'll be able to run harder and for longer, but your sprint won't improve at all.

Brilliantly put. Improving the thermal characteristics of an electrical motor improves it's continuous watt rating but not it's saturation point. I love the work you do, your actions are as inspirational as they are interesting.

 

[John in CR]

Heat is a formidable enemy, and because our motors are not specifically designed for liquid cooling there's no one stop shopping. Since heat moves relatively slowly, even motors specifically designed for liquid cooling are ineffective at heat xfer compared to ICE's due to the drastically lower temperature differentials. Maybe I'm wrong, but I don't even view the copper as the most temperature sensitive part of our motors, and instead view the magnets and their glue as having the lowest thermal limit, along with the hall wiring. This is of course talking about quality motors that don't include low spec wiring insulation or hall sensors.

The best way to eliminate heat is not to make it to begin with. Once created I view heat as a foe requiring a multi-pronged approach to keep under control, so I want to take bites out of it from as many directions as possible.

FWIW pushing our motors anywhere near saturation is a waste of battery capacity, so investing that excess weight in more motor significantly reduces heat for the same performance.

 

[fechter]

OK, out of left field, another idea I've been looking at is heat pipes.
Basic primer: https://en.wikipedia.org/wiki/Heat_pipe

Heat pipes can carry large amounts of heat over a distance using a relatively tiny piece of tubing. How to apply this to a hub motor in a practical way hasn't appeared yet, but in theory you could run a heat pipe out the axle like the phase wires and have an external heat sink. This may have some kind of practical application on a geared hub where there isn't a very good path to the outside. Even if the heat pipes just went from the stator windings to some other place inside the motor that was close to the shell it might help. A set of fins and a small fan inside the hub could work also but the fans would be problematic.

The advantages are no moving parts and they are totally sealed. These are used extensively to cool computer CPUs.

I totally agree that it would be better to not generate any heat in the first place, but present day technology doesn't give us many places we can make improvements. Someday if they make something better than copper (graphene?) that has lower resistance, that would make the cooling issues pretty much disappear.

 

[justin_le]

OK, out of left field, another idea I've been looking at is heat pipes.
Basic primer: https://en.wikipedia.org/wiki/Heat_pipe

That's already been discussed at some length earlier on this thread, have a read:

https://endless-sphere.com/forums/viewtopic.php?f=2&t=48753&p=1069962&hilit=heat+pipe#p1069962
https://endless-sphere.com/forums/viewtopic.php?f=2&t=48753&p=1031782&hilit=heat+pipe#p1031782
etc.

I ended up tossing my collection of heat pipes back to the recycling bin without ever setting up any tests. The main thing realized is that even if you DO move all the heat out of the stator through the axle area via heat pipes (which will be no small feat, research the maximum watts/cm^2 heat flux possible with reasonable deltaT's), you still need a large radiating surface outside the motor in order to get rid of that heat. Imagine having giant finned heatsinks coming off your forks that are as large as your motor side covers. By contrast the motor shell itself can already do double duty as that large radiating surface to ambient air with no extra parts or weight, and if the typically smooth and flat shell design doesn't have sufficient conductivity for a really high power application, then there is plenty of room to optimize the motor shell with fins of all kinds.

At least it sure seems a easier and cleaner to add fins to the motor body than to have a whole separate finned structure mounted elsewhere on the bike. Unless maybe you could run the heat pipes along your bicycle frame tubing and use that as the structure to sink the watts.

[edit]

Even if the heat pipes just went from the stator windings to some other place inside the motor that was close to the shell it might help. A set of fins and a small fan inside the hub could work also

That is one area where there could be merit, since the geared hub motors have a much more constrained internal geometry with the stator nested inside the spinning rotor which in turn is nested inside the shell. With heatpipes inside the stator support block you could help spread out the origin of the heat to a broader area in the motor and reduce the peak winding temps.

 

[John in CR]

I totally agree that it would be better to not generate any heat in the first place, but present day technology doesn't give us many places we can make improvements.

I disagree. Maybe not at the $80-150 FOB China price point, but based on the prices typically paid by ESers for hubbies, a solid 30%+ decrease in heat generation is there for the taking with better motors. Lower resistance by using more copper along with more and better stator steel that decrease core losses using lower slot and pole counts, so the higher rpm of a smaller wheel still results in lower operating frequency, combine for solid improvement in motor efficiency.

Good heat dissipation design will add efficiency (lower temps of the copper), and also increase power rating to offset the weight gain above.

There are also significant gains (less heat generated for similar performance) with better controller tuning and mass production of FOC sine controllers (to force their prices down to sane levels).

Then, as I've repeated for the upteen millionth time, adoption of smaller wheels for DD hubbies means less heat generated despite higher performance.

While I don't have Justin's patience or tools to prove the above through definitive testing, I still prove it through real world daily use, because the performance and temperature levels my daily rider achieves pushing my load would be impossible otherwise. If I was wealthy then I could afford production runs of sub 8kg ebike hubbies by the scooter factory I deal with using their design in a smaller size. These new motor(s) would include the inner and outer shell changes I've discussed with the factory motor tech guy to achieve better heat dissipation to compliment their superior motor design.

Forgive me for the overused description, but there's plenty of "low hanging fruit" for the taking. China doesn't make them due to average ebike cost of less than $400 in the biggest ebike market in the world. Thanks to Justin there's a real possibility that adding a few ml of FF can be added to that list. As an example of how much opportunity is out there for the taking, how is it possible that I'm still the only one I know of using something so simple as to deflect extra air flow at my hubmotor?

 

[bunya]

My interpretation... for high power applications the thermal transfer between the stator and the rotor/case can be improved by increasing windage losses.

In the design of motors the windage loss factor between the stator and rotor plays a large role in continuous power rating, if keeping the motor sealed is a priority we could try and increase the windage loss by adding alloy fins inside the motor shell. Alternatively for high continous power applications the motor could be designed more like the Mitsuba solar car or avanti electra flux motors where the stator is not encased on one side so removing heat from the stator can be done directly to ambient.

Cheers,

 

[Punx0r]

Greater windage via internal features was where this thread was headed before the subject of Ferro-fluid came up. It seems to achieve the same thing but much easier to implement (especially on a low volume/DIY/retrofit basis)

 

[Kodin]

Here's one folks haven't necessarily considered, (at least that I've read; I admit I haven't trolled the entire thread,) air gap between stator and rotor. The tighter the tolerances, (assuming all the machining is reasonably concentric to the shaft,) the less "insulating" air in the gap. Assuming it's practical, (uniform magnet thickness/mount height, rotor/stator are perfectly concentric to the shaft/bearings, could one decrease the gap to say, micrometer-level tolerances with precision potting molds or else machine the whole thing after casting? For that matter, could you simply use an iron stack that fits with similar tolerances in the magnet ring? Any disadvantages to doing this other than "get it wrong and wear down your magnets"? I'd assume you would also have to take into account material flex under load and bearing "give" when going over bumps, etc... in use. Any idea what's a reasonable gap width to shoot for?

 

[macribs]

Here's one folks haven't necessarily considered, (at least that I've read; I admit I haven't trolled the entire thread,) air gap between stator and rotor. The tighter the tolerances, (assuming all the machining is reasonably concentric to the shaft,) the less "insulating" air in the gap. Assuming it's practical, (uniform magnet thickness/mount height, rotor/stator are perfectly concentric to the shaft/bearings, could one decrease the gap to say, micrometer-level tolerances with precision potting molds or else machine the whole thing after casting? For that matter, could you simply use an iron stack that fits with similar tolerances in the magnet ring? Any disadvantages to doing this other than "get it wrong and wear down your magnets"? I'd assume you would also have to take into account material flex under load and bearing "give" when going over bumps, etc... in use. Any idea what's a reasonable gap width to shoot for?

I like the idea, but better men then I must do the math for that
I mean if you tighten in on the tolerance, when will heat and metal expansion become part of the equation? I am thinking that the gap is there for a reason. As the different material heats up and cools down inside a motor, various parts of motor probably will retract and expand on different time/degrees celsius due to different material. Suddenly the tiny micrometer gap is too tight as something inside the hub expands as it heats up?

For all I know this might not be an issue but that was the first thing that entered my mind.

 

[Kodin]

FWIW, I'd assume if you have enough heat transfer to the covers, the aluminum will expand first, meaning it may want to increase the gap as heat increases faster than the iron would. That said, I honestly don't know how it works without doing real testing and measurement.

 

[Marc S.]

Here's one folks haven't necessarily considered, (at least that I've read; I admit I haven't trolled the entire thread,) air gap between stator and rotor. The tighter the tolerances, (assuming all the machining is reasonably concentric to the shaft,) the less "insulating" air in the gap. Assuming it's practical, (uniform magnet thickness/mount height, rotor/stator are perfectly concentric to the shaft/bearings, could one decrease the gap to say, micrometer-level tolerances with precision potting molds or else machine the whole thing after casting? For that matter, could you simply use an iron stack that fits with similar tolerances in the magnet ring? Any disadvantages to doing this other than "get it wrong and wear down your magnets"? I'd assume you would also have to take into account material flex under load and bearing "give" when going over bumps, etc... in use. Any idea what's a reasonable gap width to shoot for?

I like the idea, but better men then I must do the math for that
I mean if you tighten in on the tolerance, when will heat and metal expansion become part of the equation? I am thinking that the gap is there for a reason. As the different material heats up and cools down inside a motor, various parts of motor probably will retract and expand on different time/degrees celsius due to different material. Suddenly the tiny micrometer gap is too tight as something inside the hub expands as it heats up?

For all I know this might not be an issue but that was the first thing that entered my mind.

As a rule of thump, steel 'grows' or 'shrinks' at about 0.001mm/per 100mm full Material/per 1°C Temperature difference. A 180mm iron stator should expand roughly about 0.288mm when the temperature rises from 20°C to 180°C. A steel ring will 'grow' more, since the expansion follows the circumfence, not the diameter.

If you want to have a minimal air gap, you might want to mill the inside of the rotor, to have all magnets at the same hight and centric to the axle. I was grinding some small 10mm diameter rare earth magnets on a precision grinding machine once and it was frocking difficould to measure these things with all the small chippings still sticking to it... (rumor has it they are on the ISS right now, but I wouldn't bet my bike on it)

 

[Kodin]

In comparison, how much will the aluminum rotor "grow" over the same temperature differential? I assume rings will want to expand outward, not in at all, correct? (wall pushing on itself as it expands causing all expansion to be outward in direction)

 

[Punx0r]

Tight tolerances in mass-produced assemblies becomes expensive very quickly.