Definitive Tests on the Heating and Cooling of Hub Motors
- Thread starter justin_le
- Start date
The only problem right now with water cooling our motors is that we need to replace the axle to allow for the cooling hoses and a new side cover to fit.
I was wondering if it is possible to use small water hoses with higher flow rate to be able to fit some cooling lines without modifying anything.
I think that more of us would be water cooling if we could buy a motor with these modifications done already. It really isn't all that hard to run some cooling lines and add some water blocks or copper hoses for the water cooling.
Actually it would be a lot of fun to install a water cooling system on my bike, I guess it would be all PC water cooling parts. I have plenty of room on my raptor to put custom radiators and pumps around.
I think we should contact QS motors to manufacture their cromotor type motor with a modified axle and side cover so we could easily water cool it.
Water cooling and then use the FF fluid would really do a lot.
I was wondering if it is possible to use small water hoses with higher flow rate to be able to fit some cooling lines without modifying anything.
I think that more of us would be water cooling if we could buy a motor with these modifications done already. It really isn't all that hard to run some cooling lines and add some water blocks or copper hoses for the water cooling.
Actually it would be a lot of fun to install a water cooling system on my bike, I guess it would be all PC water cooling parts. I have plenty of room on my raptor to put custom radiators and pumps around.
I think we should contact QS motors to manufacture their cromotor type motor with a modified axle and side cover so we could easily water cool it.
Water cooling and then use the FF fluid would really do a lot.
Ohbse
10 kW
Thanks again for all the hard work Justin, I have ordered one of each variety of Statorade and will provide some additional data 
I have a Crystalyte HS3540 (which has previously been oil cooled) and a Cromotor V3 in stock condition.
Guys - all the talk of insanity level hot rodding is off topic. While I am very interested and very happy to weigh in on the topic I don't think this is the best venue. It dilutes the core theme - definitive testing. Supposition and anecdotes have their place elsewhere.
I have a Crystalyte HS3540 (which has previously been oil cooled) and a Cromotor V3 in stock condition.Guys - all the talk of insanity level hot rodding is off topic. While I am very interested and very happy to weigh in on the topic I don't think this is the best venue. It dilutes the core theme - definitive testing. Supposition and anecdotes have their place elsewhere.
themadhatter106
100 W
Offroader said:
You don't need to replace the axle. Doc has done water cooling in an X5 with a modified axle.
I'd like to see more people try water cooling and plan to try myself someday. I don't see why say on a 9C you couldn't drill out the other side of the axle and run two lengths of 2mm copper tubing through the axle. Have a copper block to run the water through which bolts down to the stator. Then plumb the water out to a motorcycle radiator using flexible tubing and use an inline 12V motorcycle water pump. Even with the low flow rates that may be associated with such small tubing at the axle it should still make a big difference for continuous power handling.
I can see water cooling and ferrous fluid as being a real nice cooling setup!
Ohbse said:
Thanks Ohbse, and John earlier, for trying to keep things on tract. ES is an open forum, but the purpose of this particular topic is to stay on a certain theme for the benefit of anyone following for concrete info, and I appreciate those who've kept the contributions on the subject of the actual thermal test results and related input, ever difficult a battle that seems to be.
Earlier I had posted this:
What I wanted to experiment with here was using stationary scoops to help direct airflow through the large open holes in the motor to encourage lateral airflow in and out of the hub. I've mentioned again and again and again on this thread that the speed of air moving past the hub motor as a result of the bike moving forwards is MUCH faster than any relative rotational speeds from the hub spinning.
In the above example I folded some simple 'scoops' from aluminum sheet and had one zip-tied to the fork of the test stand, and another attached to the axle bolt. These are both things that could be attached with relative ease to an existing front or rear hub motor, and they allow for a much larger input 'capture' area than you can get with scoops that are bolted on the motor cover itself, but they're also potentially more delicate.
For a reasons of happenstance the motor core under testing here was now potted, so I wanted to repeat all the testing I had done previously with this motor core with one modification at a time. That meant running the motor under the following possible scenarios:
a) drilled holes on 2 side covers
b) with the drilled holes and scoops
c) drilled holes and scoops and Statorade
d) drilled holes and Statorade but no scoops
e) Only Statorade (holes taped up)
The plain numeric test results for thermal resistances, running the hub at ~160 watts with the shell IR temperature taken just from the steel ring and not the side covers, is as follows
Plotted graphically, things are a bit easier to see. If we start with the stock unmodified hub motor, and then add the large vent holes around the perimeter, the thermal resistance drops in half on average, reaching a 116% improvement at 40kph speeds. Adding the scoops improves things by another 23% (40kph resistance dropping from .177 to .144 deg/watt). When we then add the 5mL or Statorade Ferrofluid, the thermal resistance improves another 31% still, dropping down to .110 deg/watt.
When we compare that to the the original unmodified motor at 40kph, the net resistance drops from 0.384 to 0.110, a 3.5 TIMES improvement in thermal conductivity with the compounding effects of passive ventilation with scoops and Statorade. That is a massive total increase in heat shedding capability from the motor core, and it corresponds to almost a 90% increase in the continuous current and torque capability of the motor. So yes, for those who had been wondering there is still benefit to the ferrofluid even with a motor that is already open and air cooled. I can't speak to the overall wisdom of having the FF in an open motor as far as long term contamination etc. but I can say that not a single drop spilled out from the motor during testing with the large holes and scoops for encouraging sideways air flow.
Now, approaching this result from the other direction of first adding the Statorade, then drilling the holes, and then adding the scoops, it's a similar progression
The Statorade by itself improves things by a factor of 2, the addition of large vent holes leads to another 40-50% improvement, and scoops on top of this lead to another 20% gain.
Here it is with all the plots super imposed. Make of it what you like
If there is to be a rough conclusion from all this, it's that the addition of 5mL of ferrofluids is more or less equivalent to drilling large vent holes on both side covers of your hub motor, with both approaches roughly doubling the core to ambient conductivity, though the FF working a bit better at low speeds and Vent Holes having an edge at high speeds. The vent hole performance can be improved somewhat (~20%) with the addition of scoops to help direct air-flow through the hub, but a greater (30%) effect is had by adding FF, and the combination of all these techniques leads to a compounding benefit that is 3.5 times better than the stock hub.
Ha, even I get lost with the length of this thread!made_in_the_alps_legacy said:
Lucky for you I didn't forget about your earlier question on this topic a few pages back. I didn't answer right away because since I've changed the test setup around using the 3 discrete IR sensors I haven't had the full IR camera set up on the test chamber, but for this latest run I hooked it up again, for your viewing pleasure:
This is running with the large vent holes in the side covers. You can see that the axle itself is completely dark, and that the inner section of the hub motor plate is noticeably cooler than the outer section at the perimeter of the side plate. Don't read too much into the apparent hot spots on the motor windings, that is just an artifact of aliasing between the camera's IR sensor and the passing holes of the side cover plate.
It should be clear from this that pretty much all the heat is escaping out of the perimeter of the motor and carried by passing air flow, and that the inner section of the side cover around the axle has minimal role.
Yes for sure, it logs everything that the CA output from the datastream which includes the one temperature sensor on the CA3. But the Analogger also has two analog inputs of it's own which can be connected to two additional temperature sensors, allowing you to log 3 separate temperatures in your system. This allows you to have a separate log of the ambient temperature (quite important to the calculations) and could be interesting if you wanted say a temp probe directly on the copper and another on the stator laminations.John in CR said:
For comparative effects then you could just do a single hill climb with the blades, let the motor cool to the same starting conditions, and then repeat the exact same run without the blades. It's not really necessary to let the core get super hot, but you do need to log the full trip data in order to do a best fit to the model for getting resistance values since you won't have time on a hill to actually reach steady state.
Yeah, field weakening definitely makes the test process a lot easier, even without a wind tunnel you can ride the bike on flat ground and still have the motor heat up as though you were climbing a monster hill, and it also means that the heat generation is much more constant during a run and independent of your velocity (which isn't at all the case when hill climbing).
That would do the trick pretty well. It's a fair bit harder to extrapolate actual conductivity terms in W/K when you are measuring the total input power. I know that your controller doesn't allow field weakening, but is it possible to set a maximum phase current limit that is reached for the bulk of the hill climb? If so, we'd have a much easier time estimating the I^2R copper losses and core losses during the run since we can assume that the phase current is a known constant, and then do a best-fit model of the temperature rise data to get actual thermal conductivity values, even ballpark ones.
The other way to do the testing is to first get the motor really hot, not really caring about how, and then coast downhill while logging the temperature data as the motor cools. So long as we have a decent estimate of the heat capacity of your motor core, then the dT/dt as it cools gives us a very good number for how many watts are being shed. Given your setup this might be the more useful even if somewhat abstract approach.
made_in_the_alps_legacy
10 kW
Thxs a lot justin, those experiments must be time consuming... maybe u have some axle temperature of a closed hub with and without FF... so we make sure it would require to much efforts for few less gains to machine " melting wax pocket " in the stator arms
comments about scoops/impeller test:
At the moment, both scoops are pushing air IN on one side of the hub, maybe you could place one scoop facing downstream, so it would work as "push IN - pull OUT" aparatus forced air cooling... or even 4 scoops, 2 on one side facing upstream, 2 on the other side facing downstream...
Anyway, thxs again and looking forward for the next tests
comments about scoops/impeller test:
At the moment, both scoops are pushing air IN on one side of the hub, maybe you could place one scoop facing downstream, so it would work as "push IN - pull OUT" aparatus forced air cooling... or even 4 scoops, 2 on one side facing upstream, 2 on the other side facing downstream...
Anyway, thxs again and looking forward for the next tests
macribs
10 MW
- Joined
- Jul 22, 2014
- Messages
- 3,702
Wow. FF only is almost right on pair with vented side covers. That I would have never guessed. Glad you now have solid test data so we can choose correct cooling.
And for us living where the conditions are not kind to open side covers this is great news.
The fact that those scoops works well too, makes me wonder if the scoops would also give any benefits if used solely in pair with FF - without drilling the side covers?
A little extra cooling from the scoops when used with FF and I am sure that would be all I would need, even pushing my big behind up heavy climbs.
As those scoops are stationary maybe you could position them to direct the air stream across between then flanges? Or could it possible work just fine the way they are mounted now even without vented sidecovers?
I guess what I am asking is if you ddid tests of vented side covers + scoops + FF Vs closed side covers + FF + scoops.
And for us living where the conditions are not kind to open side covers this is great news.
The fact that those scoops works well too, makes me wonder if the scoops would also give any benefits if used solely in pair with FF - without drilling the side covers?
A little extra cooling from the scoops when used with FF and I am sure that would be all I would need, even pushing my big behind up heavy climbs.
As those scoops are stationary maybe you could position them to direct the air stream across between then flanges? Or could it possible work just fine the way they are mounted now even without vented sidecovers?
I guess what I am asking is if you ddid tests of vented side covers + scoops + FF Vs closed side covers + FF + scoops.
made_in_the_alps_legacy said:
Yes, time consuming is an understatement. I've been staying at work until 1-2am most nights for the past 2 months in order to run the wind tunnel and process data since it's just too noisy to do it in the engineering lab when all the staff are around. In the process that also meant falling behind in many other responsibilities which I need to get back to, so for now this whole round of testing has pretty much run its course, which is well summarized by the final graphs in this post https://endless-sphere.com/forums/viewtopic.php?p=1120064#p1120064
Over the holidays I'll try to set up the wind tunnel and test suite to be fully automated so that I can let it run through the test sequences unattended overnight, and then further experiments could be done more expediently in early 2016.
The axle temperature won't tell you much at all here. It's steel and a terrible conductor of heat, so if the end is cold that doesn't really tell you whether the center is hot or not. The aluminum stator support is very close to the copper winding temperatures as you saw from my potted motor test results, so if you wanted to try the melting wax then so long as it's in the motor stator support area it would pull the heat just fine. The outside end of the axle (cooled by the passing air) isn't much of an indicator of anything.
I had thought of doing that indeed, but then in the end I realized in most bike systems you have a disk rotor that blocks most of one side cover, so for external scoops it's only the non-disk side of the hub that is amenable to having a non-rotating air redirection like this.
macribs said:
Exactly. This takeaway here for me has made the whole undertaking most worthwhile, and now one of those neglected responsibilities (getting our own thru-axle hub motor design finalized and in production) can now proceed with a method that has excellent core cooling while staying sealed.
macribs said:
This could also have the opposite effect of just slowing down the air that is passing by the hub side cover, and creating a wind shadow downstream of the scoops, so that the overall convective air cooling on the side plates is worse. Remember that the scoops are not really increasing the air velocity on the surface of the side covers, which is what it would need to do for better shell->ambient heat flow. I think the previously discussed technique of having additional fins on the rotor ring itself would be the best way to further FF hub while keeping it sealed.
...Next year...
Samd
10 MW
I'm glad that you've clarified the opportunity for venturi cooling versus active/internal cooling methods.
Some sort of radial venturi system would be cool. Pardon the pun.
Some sort of radial venturi system would be cool. Pardon the pun.
Those are some great test results. Beats the snot out of guessing.
speedmd
10 MW
Link is possibly some help for getting realistic estimates for what to expect from external cooling fins.
http://www.electronics-cooling.com/2003/05/how-much-heat-can-be-extracted-from-a-heat-sink/
http://www.electronics-cooling.com/2003/05/how-much-heat-can-be-extracted-from-a-heat-sink/
liveforphysics
100 TW
Thank you for your hard work!
Excellent to see easy mods to the motor buy the continous power of two motors.
Excellent to see easy mods to the motor buy the continous power of two motors.
John in CR
100 TW
justin_le said:
I can set battery and phase current limits, though I'm not sure how trustworthy the phase limits are with no way for me to measure them. Also, the nearby climbs are all very curvy enough to make constant speed impossible.
justin_le said:
Great idea. I hadn't thought of that direction, but I really like it...well except the "really hot" part, but I get to have some fun riding hard and heating the stator up to 100°C, which is hot to me regardless of how much more the motor can take. My motor is in Miles' spreadsheet, so we'll be able to compute the iron losses added during the decent using rpm data. I just need to choose a long shallow grade descent that won't need regen and preferably no braking at all with the motor's disc brake.
It will be a while, but I figured out how to easily add multiple perimeter heat sink discs to my mid size motor with the aluminum shell. If the FF proves durable and useful for 1000+rpm motors in the 10 pole pair range, then I'll be able to test the effectiveness of different perimeter heat sink strategies.
John in CR said:
Well there's no need to necessarily get the motor really hot. It just helps improve the resolution a bit since you'll get more temperature drop during the downhill, but the data would likely be just as good starting at 60°C or 80°C, especially since you have 0.1 degree resolution with the CA3. I think if you just climbed the hill relatively slowly you should have no problem reaching the target temp at the top of the hill then being ready for the coast down.
Assuming that your motor is on the rear, then yes exactly, use your front mechanical brakes to keep a more or less constant speed on the downhill. You could get some very good data from this experiment in just a single outing. Climb a relatively steep hill slowly to get the motor core warm, then turn around and coast downhill using the mechanical brakes to keep a constant speed. Then turn around and slowly climb the hill again to get the motor warm, and repeat (either coasting at a different speed, or at the same speed but with a different cooling mod on the hub). It looks like you'd get enough temperature differentials even with just a 300-500m or so elevation change which would allow several trips up and down per hour.
With the logged data, you'll have a plot of the motor core temperature vs. time. What I usually do is create an adjacent column in the spreadsheet that has the calculated motor core temperature based on the first order model. So the Temperature at n+1 seconds is based on the Temperature at n seconds with the following relationship
T(n+1) = T
+ (Core Loss + [T(ambient) - T]/K )/ Heat Capacity Where 'K' is your core to ambient heat conductivity, and the Heat Capacity is an effective heat capacity of the core plus part (say 30%) of the shell's heat conductivity. You then adjust the value of 'K' until the temperature curve predicted by the model lines up neatly with the measured temperature curve, and the best fit value becomes the thermal conductivity term for that speed.
We're actually waiting for a non-rainy day here in Vancouver to do some field testing and video footage of a mountain climb with and without the Statorade in a hub, for those who want to see results on actual bike rides rather than lab data. I hadn't really thought of logging the downhill bits but this gives me some impetus to do that as an example of using the method, and then compare how this coasting/cooldown derived conductivity terms match up with the measured wind tunnel results (hopefully very closely!)
John in CR
100 TW
The thing I like about the coast down method is being able to calculate how much heat is being dissipated. I don't know how much heat is generated, so I don't know how much of my low motor temp results is attributable to the high efficiency motor and controller tuning, and how much is attributable to my approach to ventilation.
Without setting current limits quite low it's all but impossible for me to run a continuous current for accurate direct comparisons. The coast down cooling method changes everything. A after weighing the different parts, I'll know with reasonable accuracy how much heat the motor holds. Then I'll know how much heat is dissipated, which makes all kinds of analysis possible... not only changes in cooling efforts, but once I can quantify the heat dissipated based on rpm then the Analogger's data will help me look at heat generation and real world efficiency.
It will help in a big way to plan some very long runs that include mountain pass climbs. Right now I only have a gut feel for what speed to attack the mountains, and the temp gauge with alarm is just a safety protection that I rarely turn on. Climbs long enough to create continuous conditions are a completely different ball game than my around town riding. Only the few kilometer 20% grade climb to the wind generator farm along the ridge above town comes close.
Without setting current limits quite low it's all but impossible for me to run a continuous current for accurate direct comparisons. The coast down cooling method changes everything. A after weighing the different parts, I'll know with reasonable accuracy how much heat the motor holds. Then I'll know how much heat is dissipated, which makes all kinds of analysis possible... not only changes in cooling efforts, but once I can quantify the heat dissipated based on rpm then the Analogger's data will help me look at heat generation and real world efficiency.
It will help in a big way to plan some very long runs that include mountain pass climbs. Right now I only have a gut feel for what speed to attack the mountains, and the temp gauge with alarm is just a safety protection that I rarely turn on. Climbs long enough to create continuous conditions are a completely different ball game than my around town riding. Only the few kilometer 20% grade climb to the wind generator farm along the ridge above town comes close.
justin_le said:
Do we get video of a hub motor smoking?
Estimating the heat capacity of a motor might be a bit tricky, since different parts will heat/cool at different rates. If you knew the weight of copper and iron, you might get reasonably close. Correlating the heat capacity of a motor to it's weight on a carefully measured setup should allow a good approximation on motors of similar construction.
fechter said:
Exactly. Please read THIS POST Titled "On The Motor Heat Capacities" where we already measured and tabulated the heat capacity of a large number of direct drive motor cores and motor shells, from which you can see if you approximate 0.5 J/K/g for the motor core, and 0.6 J/K/g for the shell, you'll be as good as necessary:
https://endless-sphere.com/forums/viewtopic.php?p=1022602#p1022602
John in CR
100 TW
Thanks for reminding me you did all those specific heat measurements, though your table does reinforce my use of educated guesses, since .5J/g K was ballpark estimate for my stator considering the specific heats of aluminum, steel and copper. We'll see how close I am if I can get the separate weights from the factory for the alloy center support, axle, and unwound stator assembly of my motors. With those I can get the copper amount using actual weight, but I will have to estimate the plastics.
Even if the .5 is off by a few %, I think uneven stator temperature between the perimeter to the axle will cause a greater error, as will differences in ambient conditions even at the same ambient temp reading. I'm not concerned about being too precise about the heat number, since I'm after definitive comparisons which using the same bike and motor will allow. All I need is long enough descents and maintaining the same speeds.
The actual heat number is just very interesting info that's been a big unknown before, and a reasonable estimate of heat dissipated will be like gold for me. It opens the door for all kinds of good stuff to capture. Take regen for example, the CA tells us how much goes back into the battery, and while we know it makes heat in the motor, how much has been an unknown.
Even if the .5 is off by a few %, I think uneven stator temperature between the perimeter to the axle will cause a greater error, as will differences in ambient conditions even at the same ambient temp reading. I'm not concerned about being too precise about the heat number, since I'm after definitive comparisons which using the same bike and motor will allow. All I need is long enough descents and maintaining the same speeds.
The actual heat number is just very interesting info that's been a big unknown before, and a reasonable estimate of heat dissipated will be like gold for me. It opens the door for all kinds of good stuff to capture. Take regen for example, the CA tells us how much goes back into the battery, and while we know it makes heat in the motor, how much has been an unknown.
One thing is the heat capacity of the core won't change, so comparing two cooling methods on the same motor will still be valid even if the capacity estimate is off.
thanks for the nice summary of all the different tests. about the effect of FF to no-load losses: it was already explained somewhere, but having the graph in the summary would be great.
i know it doesn't belong to this thread, but it may be of interest for some:
Dear Martin,
sorry for the deley.
the magent for 45H-3000W motor is 38H.
have a nice day.
Best Regards
Snow
i know it doesn't belong to this thread, but it may be of interest for some:
Dear Martin,
sorry for the deley.
the magent for 45H-3000W motor is 38H.
have a nice day.
Best Regards
Snow
Maybe It's already in the thread.
I had FF in my car glove box for 18 months before showing it to my sister kids.
some of it did dry around the edges of the small container.
I dont know what is the solvent but it will eventually evaporate.
I would use it on a closed side cover
How did turn out the experiments with Boeshield T-9 to protect the stator from rusting in a open environnement ?
Thanks
I had FF in my car glove box for 18 months before showing it to my sister kids.
some of it did dry around the edges of the small container.
I dont know what is the solvent but it will eventually evaporate.
I would use it on a closed side cover
How did turn out the experiments with Boeshield T-9 to protect the stator from rusting in a open environnement ?
Thanks
I might not have the W/Degree data, but hopefully this is still allowed.justin_le said:
I did some testing of my fan cooling setup on the weekend up a very steep/rough hill.
Details and video here:
https://endless-sphere.com/forums/viewtopic.php?f=30&t=56965&p=1123565#p1123565
I would like to share here what I think are my 'Definitive' test results:
I hope to do further testing under normal off-road riding conditions including many ups/downs, etc. to better gauge the normal usage performance difference. I can arbitrarily see a difference, but a proper test would still be nice.
Cheers
