So, time to give all you waiting folks an update. I got a lot of wind tunnel tests done the last few weeks, and things are verrry interesting. First off, after opening up the hub motor to clean it out from this first run with ferrofluid I noticed something that could be telling. Of the ferrofluid that didn't fly out from the unsealed gap in the side covers, well most of it stayed in these perfect ridges right at the junction between each pair of magnets, where naturally the field gradient is the strongest:
This immediately suggests that the reason why so little ferrofluid seems required to conduct the heat away. If it concentrates just as small ridges between each magnet, then these ridges could act like "thermal wipers" as they circle around the stator core, scraping heat from the surface while presenting very little viscous drag to the system since the actual fluid contact area is quite small. Until we get a camera in the air gap it'd be hard to say that's exactly what's going on, but it seems pretty plausible and certainly paints a pretty picture, kinda like a wave soldering action.
On the rest of the motor, there was obviously ferrofluid stains on the stator iron and on the side cover flanges where it initially spilled out of, but nothing at all present further up the side plates near the axle, bearings, wires etc. confirming that all the leak was from the poorly sealed side plates.
Next up I'll want to do the reverse, rather than overfilling the air gap and letting it leak we'll keep adding say 1-2mL at a time and then measuring the thermal transfer coefficient, and find out at what point you get diminishing or no returns.
So this test took a little while. After emptying out and cleaning the motor, I was careful to properly seal the side plates and viewing window this time with silicone and then drilled a small ~3mm hole in the side plate for the fluid injection. This was applied with a syringe, first 1mL at a time up to 6mL, and then in 2mL increments until I ran out.
I did all the heat transfer tests with the motor spinning at 200rpm and the wind tunnel at ~20kph, with 150-160 watts of heat being generated in the stator core from a 30A field weakening current from the controller.
Here's the raw numeric data along with the calculated conductivity values. The shell to ambient varies a bit because in some cases I had the back end of the wind tunnel blocked by an obstruction so the internal air flow was slower, but that doesn't matter too much because the numbers of real interest here are the core to shell values, not the shell to ambient. With no fluid the conductivity is 3 °C/watt, adding 1mL doesn't change this at all, 2mL shows the slightest increase (3.3 °C/watt), and then from 3 to 4mL it jumps right up to 6 °C/watt and by 6mL is plateaued at about 7.
The results plotted in a graph are pure gold and tell the story perfectly.
There's really no need for more than 5mL of fluid in here to get the full thermal transfer effect. After 6, it's not just diminishing returns, it's no extra returns. But what you do get above 6mL of fluid is extra drag on the motor. In conjunction with all the thermal measurements I also ran the no-load drag test with each fluid fill, typically after the wind tunnel tests so the motor cores were still fairly warm, around 50oC. Here is the resulting motor drag at both 200rpm and 400rpm.
I think most of the variation can be attributed to the lack of temperature consistency when the measurements were made, but it's apparent that from 4mL and below there is no change in the drag, from 5mL onwards we start to see it increase, and it keeps on increasing right up to the 16mL measurement point. From this data we can easily conclude that there is a definite "sweet spot" for the fluid fill, in this case with MXUS motor it's around 5-6mL, which as allows for full thermal transfer with almost no extra motor drag. If you put more in, say 10mL, you won't see any increase it heat dissipation but the rolling resistance of the hub is increased by like 25%.
In order to find this ideal fill on other hubs the easiest approach would be to look at your no load current draw, then add the fluid in small increments at a time until you notice the slightest increase in your no-load amperage. At that point, stop.
Also, it's interesting to look at the thermal camera IR images of the side cover plates. In the first 3 tests from 0-2mL, the side plate temperatures were quite uniform, with the spot closest to the axle (Spot1) being slightly warmer.
However, with the ferrofluid above 4mL, then there was always a clear temperature gradient with the outside of the rotor and side plate (spot 3) being 2-3 degrees hotter than the middle section, just as we'd expect if the heat flux was coming to the motor shell from the perimeter magnet ring and then working it's way out from there.
flathill wrote:also note the viscosity lowers as the fluid warms up. there is another "loosening" effect that is not temp related (less particles clumping) after being "worked"
make sure to do the drag test on a "cold" motor as-found after a fresh fill and then repeat the test after the fluid has been worked/warmed
Indeed, I could see this even with 5-10 degree temperature variations, so in order to better characterize this effect I ran the motor full throttle at 60V, first from a cold stator until it self heated, and then after heating the stator to 75oC and recorded the no load current draw as it cooled down. Here's the X-Y scatter plot of that. Going from room temp to 70 °C results in a 30% decrease in the motor drag. Since a lot of that drag is from core losses rather than the fluid, then the % decrease in viscosity from the fluid itself would be a lot higher.
macribs wrote:Maybe you can do another run with sealant on the side covers to see if the ferro fluid is forced outside the hub via axle/wires/bearings?
I think it has been mentioned once or twice that oil filled hub will create internal pressure inside motor that will keep pushing fluid outside the motor - I wonder if the same goes for ferro fluid or will the magnets keep oil in place so much that leaking is not a problem with proper sealed side covers?
So here's the best part, after a week's worth of testing and running that motor again and again in the wind tunnel test chamber with the side plate flanges sealed (but the injection hole left wide open), there wasn't a single drop of leaked ferrofluid anywhere to be seen.
justin_le wrote:We'll also do a test with conventional ATF oil fill which splashes on the copper windings.
justin_le wrote:And finally we'll try to pot the stator in a thermally conductive resin and see how well that improves the ferrofluid approach by better coupling the windings to the iron stator.
Expect all this to take a couple months.