Definitive Tests on the Heating and Cooling of Hub Motors

Bullfrog said:
I didn't install a vent because I was concerned about contamination entering the motor thru the vent since it will breathe in and out depending on the motor temperature and I only ride off road where it is dirty and dusty.
There's info about vents and devices for them to prevent ingress of stuff here in this thread; I don't remember what they're called though.
 
Thanks...I pretty thoroughly investigated the options, and yes there is a lot of good info on this thread as well as other places in this forum. Another thing I didn't mention is the spave available between my motor and frame, the allen head bolts barely clear the frame...just another factor that kept me from installing a vent.

I never intend for any of my comments to be negative in any way...I appreciate every thought, suggestion, idea, etc. that everyone posts. This forum is sort of like a brainstorming session and nobody should be criticized in any way for their idea...they are all helpful :D .
 
made_in_the_alps_legacy said:
justin_le said:
We had someone drop off an Inboard electric skateboard with damaged controller that provided a good candidate motor for these experiments. I put tape over the 3 vent holes that are normally on the outside of the hub.
are the measurements w. 0ml statorade w. or w.o the vent holes taped ? (those kind of hubs exist with " venting - forced aerothermal heat transfer" - and w.o ...) thxs

As always very astute line of questioning from you. I actually did a pretty detailed test on the Skateboard hub motors since this is one application where the motors definitely have overheating issues, and with the much smaller mass of a skateboard hub stator there isn't nearly as much time bought from the stator's heat capacity. However, on a skateboard motor the rotor is insulated by a rubber wheel, so it doesn't have the same direct pathway to air cooling as an ebike hub motor and I wasn't sure if Statorade would have enough of an effect to be worthwhile.

I did most of the testing on this inboard motor, both with and without tape covering the vent holes
Skateboard Motors, with and without Tape.jpg

Here is the net conductivity results with the open vent holes, with the vent holes taped, and with them taped plus 3mL of Statorade. At low speeds, the vented holes are only marginally better than having the side side cover blocked and not nearly as good as just Statoarde. At high speeds though, the performance with the open vent holes improves and basically reaches par with the Statorade motor at ~38 kph. Beyond that I would expect the open holes to outperform.

Inboard Motor Results Tape vs Vented.jpg

Still, this does show that over the typical speed range used in an electric skateboard, you'll have better thermal performance with a sealed motor that has Statorade in it than you would with a motor that has cooling holes on the outside side plate. But the overall effect is much lower than you get with ebike Hub Motors, and the main reason for that is presumably the rubber wheel interfering with direct heat flow off the rotor.

It's worth pointing out that in this graph I am plotting the net conductivity from stator to ambient, while in my previous post that was just the core to shell conduction which is useful for certain aspects of analysis doesn't really tell the full picture. Unlike an ebike hub motor, the "shell" temperature on a skateboard wheel is not as easy to measure since it's under the rubber tire.
 
justin_le said:
But when I do the data analysis for conductivity, we're not seeing ANY change at all so far.

file.php


It's been hovering right around 4.75 W/K for the first 24 days. That's 29,000 km.

I've now got another 12 days of test data here since posting this.
MXUS Evap Test at 120oC, 9mL, Raw Data, days 24-35.jpg

It seemed from this raw test data like the average input power to maintain 120oC was decreasing a little. And sure enough when I do the thermal computation on this data and add it to the previous results to day 24, we're starting to see a very slight decline in conduction. Now w'ere averaging about 4.65 W/K from stator to shell down from an initial 4.75 W/K (versus 2.33 W/K with no Statorade).


MXUS Evap Test at 120oC, 9mL, Day 36.jpg

This could be a long time running before it drops more substantially.
 
Anybody want to see some good info on electric motors?

Check out this presentation by Justin back in 2011: https://www.youtube.com/watch?time_continue=7&v=gwlbAJLzI_w

Wish I had seen it before I did my testing with ATF, I probably wouldn't have done the test...naaaah, not really...I am too hard headed and would still try it but I would have been a little better informed.
 
justin_le said:
made_in_the_alps_legacy said:
are the measurements w. 0ml statorade w. or w.o the vent holes taped ? (those kind of hubs exist with " venting - forced aerothermal heat transfer" - and w.o ...) thxs

As always very astute line of questioning from you. I actually did a pretty detailed test on the Skateboard hub motors since this is one application where the motors definitely have overheating issues

Here's some tests from another skateboard motor, the Jacob's hub. A lot of these skateboard motors seem to have a common stator structure with 24 teeth and 28 magnets, but in this particular version the magnets themselves are thicker than I'm used to seeing.

Jacobs Sk8 Hub Opened.jpg

This motor has smaller size vent holes the Inboard skateboard wheel I posted about previously, although it has them present on both the inner and outer side plates while Inboard only had vents on the outer cover. I did a similar thermal characterization, only this time I did it with and without the vent holes covered, and also with and without 2mL of Statorade added

Jacob Sk8 Hub in Tunnel.jpg


Here are the total thermal conductivity results.
JacobSK8_VentsStatorade.jpg


The slope of the conductivity vs speed curve is shallower with the vent holes covered versus the having the vent holes open. About 0.015 (W/K)/kph vs 0.03 (W/K)/kph. In both cases of the vents being open or taped up, the addition of 2mL Statorade to the motors seems to add a steady 0.2 W/K improvement to the conductivity over the entire speed range.

In this motor, it's definitely better at most speeds to have vents and no Statorade, versus Statorade and no Vents. But if you have both the vent holes and Statorade the effect is best, just not sure how long the fluid would last in that case.


I put this motor on the simulator too in the natural state, with Statorade and the vents blocked, and also with Statorade and the vents left open, in case anyone wants to play with the results and look at the overheat times and steady state temperatures. The difference is not as pronounced as with ebike hubs, but is definitely there.
Jacob Sk8 on Sim.jpg
 
Last summer I posted my first quantitative tests on hubsinks with Statorade on a Crystalyte H Motor
https://endless-sphere.com/forums/viewtopic.php?p=1308270#p1308270

I've done a number of additional tests since then and owe it to Sketch to finally share these results!

First, one thing that occasionally comes up is people talking about using Hubsinks or other heatsinks glued to the side of a motor shell without any Statorade or fluid in the motor. Related to the previous post on the H3540 motor, I also did a run with the hubsinks and no Statorade. The result is as expected, just a very slight improvement (Green plot) over nothing at all (blue plot), and a very long shot from matching Statorade without Hubsinks (red plot).

H3540 Conductivity with Hubsinks.jpg


We started carrying the 30mm stator MXUS motors earlier this year and I got thsoe thermally modelled too.
MXUS with Hubsinks.jpg

For core the core to shell conductivity, Statorade has a large effect, while Hubsinks do basically nothing additional:

MXUS_Hubsinks_Core.jpg

Meanwhile, for the shell to ambient conductivity, Hubsinks have a huge effect while Statorade does almost nothing:

MXUS_Hubsinks_Shell.jpg

And finally, for the net total conductivity when both of these results are combined we have a set of results that looks like this, with Hubsinks increasing the conductivity of a Statorade only hub by about 0.6-0.7 W/K on average

MXUS_Hubsinks_Total.jpg
 
Wow, that´s really interesting. I didn´t expect the hubsinks to do nothing if they are not connected by ff.. Thanks Justin! You brought some light to the ebike world - once again :D
 
I had a predetermined that hub sinks would improve the cooling much more then that. Its now apparent that the shell gets adequate airflow by its spin which makes good sense :)
 
eCue said:
I had a predetermined that hub sinks would improve the cooling much more then that.

I was actually expecting more of an effect too, but on further thought does make sense. The fins do improve the conduction from the motor shell to ambient air by solid ~50-60% or so as you see in this graph here:
https://endless-sphere.com/forums/download/file.php?id=231342

But even with Statorade or other fluid in the hub, the larger barrier to heat flow is still from the motor core to the motor shell. For instance at 300 rpm with the MXUS motor core to shell is 6 watts/K, while shell to ambient is is 11 W/K (without fins). There's nearly twice as much thermal resistance inside the core. That means every % boost in shell to ambient conductivity from fins, external forced air flow etc. results in just about 1/3rd that much percent increase in the total conductivity from core to ambient. And hence that 50-60% boost on surface conduction gets reduced to 15-20% for the net total effect.

Here's a few more test curves with Hubsinks. First the large and powerful Cromotor. When we look at the shell to ambient conduction it tells the same story of ~50-60% improvement

Cromotor Shell Ambient.jpg

And then when we look a the net total conductivity, it's more in the 20% realm

Cromotor Total Conductivity.jpg


My first wind tunnel dyno test on Hubsinks was with a generic 9C hub and the results here were not as good as the others. But on removing the hubsinks afterwords I realized that I hadn't put enough thermal grease under them as there were only a few patches of contact between the sinks and the motor due irregularities in the motor finish. So I would expect better results (similar to the H, Cromotor, and MXUS) if the fins were more properly mated.

9C_Total_Hubsinks.jpg

DasDouble said:
Wow, that´s really interesting. I didn´t expect the hubsinks to do nothing if they are not connected by ff.. Thanks Justin! You brought some light to the ebike world - once again :D

Yeah hopefully seeing data like this saves some people the trouble. Sketch has made it clear again and again that the hubsinks really require a fluid bridge (ferrofluids or ATF) in order to work and this actual test data should drive that point home.
 
Thanks for the info :)

I want to mention that the hubsink fin length has me freaked out a bit due to air resistance at speed.Im basically a full time efficiency nut and am curious if the fins need to be so long ? More so if shortening them by say 1/2 would or would not raise the motors internal temperature
 
Hi, Justin. Respect to you for your analysis and explanations. This year I added 5 ml of statoraide + hubsinks, now ambient is 8C and motor is max 92C . Could you please explain why you adding 18 ml of statoraide? My motor is QS205 and max P is around 8 kW. My max speed is about 70-80 km/h. Do I need to add extra statoraide for the better performance? Thank you.
 
eCue said:
I want to mention that the hubsink fin length has me freaked out a bit due to air resistance at speed.

It would be easy to quantify the effect of increased air drag, at least to a first order, by just looking at the no load current draw of your motor at full throttle both with and without the hubsink fins attached. That will give a good idea of the additional losses in watts that you could expect, it's not exactly the same as the extra air drag from forward motion but it should get you in the ballpark.

basically a full time efficiency nut and am curious if the fins need to be so long ? More so if shortening them by say 1/2 would or would not raise the motors internal temperature

When we were doing the initial enclosure development for the Satiator we were playing around with a finned design to to improve the heat conductivity. I wanted them short for aesthetics, like 1cm tall, but it turned out in doing tests that this was barely any better than a flat surface (which is what we went with in the end, see this post). To get good convective cooling through heatsink fins they needed to be surprisingly long, like over an inch.

In the case of hubsinks, there is forced airflow as a result of forwards vehicle movement and rotation so the situation is not exactly the same. I'd be curious of Sketch wants to chime in on why this initial length was chosen, if it was done from balancing various tradeoffs or just a guess of what felt right?

miunchy said:
Could you please explain why you adding 18 ml of statoraide?

That's easy. In the sequence of doing the tests I first do the incremental addition of Statorade 1 or 2 mL at a time until it's totally saturated to get the curve of conductivity vs. statorade fill levels. Then once I'm ready to move onto the experiments with hubsinks, the hub already has like 16 or 18mL of Statorade in it. So I just roll with that rather than opening up the motor, wiping it dry, resealing the side plates, and then adding a fresh quantity.

When you have QTY's like 5mL you'll get a reduced effect as the wheel spins at higher RPMs (say more than 300 rpm ~ 40 kph). Normally this is fine since most motor heating occurs with low speed high torque travel, but if you did want maximum cooling even at high speeds then you could add a few more mL. For an 8KW peak power hub I don't think you care about additional rolling drag, I'd go with more like 9 or 10mL in this case.
 
To me the data screams the importance of Statorade to complete the thermal circuit...without it the hub sinks are not of much use.

Hub sinks...has anyone considered turning the fins about 70-75 degrees from their current orientation? Attach the pieces the same way but turn the fins on the base. IMO with the current fin orientation will push air away from the hub and create a big dead zone as far as air flow goes and if you don't move any air across the fins then you are not optimizing the convective heat transfer. What I mean is 15-20 degrees off of being parallel to the direction the bike is moving.

Here is a link to a related study that looked at the optimum angle for louvers in a shell and tube heat exchanger...should be somewhat relevant: http://www.academicjournals.org/article/article1380627619_Jang%20and%20Tsai%20(1).pdf Haven't calculated the Reynolds number for air across the fins on a bike but my bet is that it is relatively low.

Another data point is that cars tilt the radiator back about 15 degrees to optimize heat transfer for air passing thru the radiator and across the surfaces...just another corollary.

Copper sinks would work better too but may be cost prohibitive...copper is also softer and gets bent easier. The thermal conductivity of copper is almost twice that of aluminum so it depends on what you are willing to spend and the application.

So many questions...so little time :D .
 
I like that angled heatsink fin idea with shorter heatsink fins so its not running too strongly like a fan on top of the motor :D
 
Browsing via mobile phone so hard to see the graphs clearly. Core to shell seems to be the bottleneck. Would adding yet another hubsink side by side with the first one make a worthwhile difference? I mean dual fins double the fin to air ratio. Could you maybe run a test? As the net seems to be 15-20% improvement from single hubsink even a 5-10% spike in heat shredding ability would go a long way for the added cost of dual hubsinks.
 
eCue said:
Thanks for the info :)

I want to mention that the hubsink fin length has me freaked out a bit due to air resistance at speed.Im basically a full time efficiency nut and am curious if the fins need to be so long ? More so if shortening them by say 1/2 would or would not raise the motors internal temperature

I gotta figure the fins are in pretty turbulent air to begin with - especially with a rear mounted hub. You have the front wheel, downtube, seat tube and then the rear tire and rim disrupting the air flow before it gets to the hub. And I'd bet the cyclists legs interfere quite a bit also. In other words, lot-o-turbulence. So I'd be surprised if they affect air drag all that much. Further, in situations where people are using motors that are pulling enough power to require this kind of improved cooling, it is less likely that small gains in air drag efficiency will be relevant to the end use.
 
justin_le said:
In this motor, it's definitely better at most speeds to have vents and no Statorade, versus Statorade and no Vents. But if you have both the vent holes and Statorade the effect is best, just not sure how long the fluid would last in that case.
very interresting finding, thxs a lot for the quantity and quality of your work in ur answers.
 
One thing I've noticed from actual riding experience is the air speed makes a huge difference in the heat dissipation. If I am climbing a really steep off road trail going about 5-6 mph, the motor will overheat pretty quickly. Using the same power level on the street where the speed is over 15mph, it never overheats. Part of this is due to the motor being less efficient at the lower speed, but I can tell from how fast it cools on the downhill stretches that the air speed is a critical factor.

Somebody (sorry, I forget who) mounted a fan on the swing arm aimed at the side of the motor to increase the air flow. I might give this a try and see how it does with the low speed climbs. Of course having a passive system like the hub sinks has a lot of advantages, but the fan is easier to implement and cheaper.
 
wturber said:
eCue said:
Thanks for the info :)

I want to mention that the hubsink fin length has me freaked out a bit due to air resistance at speed.Im basically a full time efficiency nut and am curious if the fins need to be so long ? More so if shortening them by say 1/2 would or would not raise the motors internal temperature

I gotta figure the fins are in pretty turbulent air to begin with - especially with a rear mounted hub. You have the front wheel, downtube, seat tube and then the rear tire and rim disrupting the air flow before it gets to the hub. And I'd bet the cyclists legs interfere quite a bit also. In other words, lot-o-turbulence. So I'd be surprised if they affect air drag all that much. Further, in situations where people are using motors that are pulling enough power to require this kind of improved cooling, it is less likely that small gains in air drag efficiency will be relevant to the end use.


That makes good sense but it does mean there is not room for improvement

. I was/am thinking the fins could be shortened without raising the interior temp because of differences in heat transfer coefficient of the shell and motor. It looks to me like there is room for improvement but not much :D I deduce the fins draw about 5 watts as no ride is free tweaking the fin length and blade angle might reduce it to 2 watts.
 
fechter said:
Part of this is due to the motor being less efficient at the lower speed

??? If you're pumping the controller's current limit into a motor, it doesn't matter at what point on the efficiency curve the motor is operating, does it? ie The copper losses are the same? If you add in the iron losses, then heat created goes up despite the increase in efficiency, though as you noted it's dissipating heat faster.
 
If this heatsink works as good as it looks it may well be the ultimate sink for the job. Low drag low weight fast reacting and hub bling at the same time.
Google patents - https://patents.google.com/patent/US7399919

US07399919-20080715-D00000.png


If it works it would be nice to have it on the ebike store shelves for the DD ebikers easy access right next to the Statorate.

US20080282542A1-20081120-D00000.png



Below is not for this application but might be useful in future projects as it can be soldered in and out of cases with the flexible design.

flex7.jpg


Looks more useful then a heatsink inside the case like is often done
 
John in CR said:
fechter said:
Part of this is due to the motor being less efficient at the lower speed

??? If you're pumping the controller's current limit into a motor, it doesn't matter at what point on the efficiency curve the motor is operating, does it? ie The copper losses are the same? If you add in the iron losses, then heat created goes up despite the increase in efficiency, though as you noted it's dissipating heat faster.

Yes, you are correct.


A flexible heat sink would have a lot of advantages if it performs OK.

Here is another style that is just folded aluminum sheet:

Making it stay on the motor would be the trick here. Silicone glue might be the most practical. If there was a way to string a piece of aircraft cable through the fins it could possibly just clamp on.

Flexible-Heat-Sink-D75.jpg
 
That reminds me a lot of the jet boil pots heatsink. Might be able to jb weld it on held it place with zip-ties or bands under pressure until its dry

Jetboil+Sol+Ti+HE+--+Good.jpg
 
fechter said:
Making it stay on the motor would be the trick here.

"stainless steel" ziptie or "hose clamp" around the heatsink fins. Can notch the fin material thru the center area to ensure centered-seating of the tie or clamp.
 
Back
Top