Measuring airflow to optimize cooling of hub motor

hias9

1 kW
Joined
Jul 11, 2018
Messages
422
I am looking for a way to measure air velocity on a certain spot (like somewhere on the hub motor/hubsinks) to optimize cooling.

My plan is to make sheets that direct air to the hub motor to improve cooling at higher speeds.
 
Not really. I would need some kind of sensor I could mount on the hub motor or hubsinks and it should log data. I did not find anything that would do the job.
 
Well, since the point of the airflow is to cool the motor, why not set up the system as a "wind tunnel" type experiment, such that the motor is loaded enough to cause sufficient heating to match your use case, and then use an FLIR or similar to view and record the heat on the entire unit as it spins?

Then alter airflow and watch / record the heat changes.

When the heat is at the desired / acceptable level, then you have the optimal positions for the airflow deflectors.


If you must instead do the testing as live rides, you can instrument the hub externally with glued-on thermal sensors and one of the various arduino / etc setups that allows you to record those (if there isn't one exactly like you need you'd have to do some coding or have someone do that for you); it would have to be cheaper than finding a battery powered test-equipment-company type of system that is small enough to fit in the spoke area of the wheel, that would do this.

Something similar can measure stator temperatures if necessary, but it sounds like you only need to know about the effect of airflow on the exterior.



Either way, it would be more effective to directly measure the effect you want the airflow to have, than to measure the actual airflow itself.
 
Like amberwolf said, if cooling is the point, measure cooling instead of measuring airflow.
 
I would try to solve this task with system like a pitot tube: https://en.wikipedia.org/wiki/Pitot_tube
You can buy such modules here, for example:
https://www.banggood.com/de/MPXV7002DP-APM2_5-APM2_52-Airspeed-Sensor-Differential-Pressure-Controller-p-1179084.html?rmmds=detail-left-hotproducts__7&cur_warehouse=CN
https://www.banggood.com/de/PT60-Tube-Air-Speed-Meter-Sensor-Kit-Differential-for-Pixhawk-APM-PX4-Flight-Controller-RC-Airplane-p-1371068.html?rmmds=detail-left-hotproducts__2&ID=510170&cur_warehouse=CN
https://www.banggood.com/de/Matek-Systems-Analog-Airspeed-Sensor-ASPD-7002-Flight-Controller-for-RC-Airplane-FIxed-Wing-p-1562710.html?rmmds=detail-left-hotproducts__3&cur_warehouse=CN

I think, the delivered probe is too large for your application, use simple plastic tubes instead.

As you don't climb up to high altitudes and we don't expect large variations in temperature during the measure cycle, you may cover the "Ps" connector with a short (e.g. 4 inch long) airtight tube. (you have to reset to zero the equipment then in front of every test run)

Lead the "Pt" tube (length up to several meters) to the point you want to check the airflow. Note: the air should meet the end wall of tube in a perpendicular way.

Calculation:
(sorry for the metric units, I'm not familiar with imperial units)

v = airspeed [m /s]
pt = total pressure [Pa = N/m²]
ps = static pressure [Pa = N/m²]
roh = densitiy of air (depending on altitude, humidity and temperature) [kg/m³]

max. range of sensor:

pitot.jpg
 
Stop overloading your motor.
 
If you could duct the flow ( all the air you wish to measure) you could use a typical MAF (outpout) to know how many CFM go through the duct. Lol. Not easy they get complicated, much more complicated than a typical Baro sensor.... But yea without using P=VrT and either way its complicated.

I am looking at a kPa digital sensor right now and it isnt accurate enough even....a reason for the complexity of MAF sensors....

Pressure moves fast. Really fast.

PV = nRT, you know, all that scientific stuff..... whatever that jazz is. Scientist stuff. Lol.
 
hias9 said:
Not really. I would need some kind of sensor I could mount on the hub motor or hubsinks and it should log data. I did not find anything that would do the job.
If you have to use a sensor to measure velocity, I suggest MPXV7002. It is sensitive enough (+/-2kPa), and cheap enough (~$20) to do what you want. Make your own pitot/static tube.

If I was going to do it, I would make about 5 manometer/pitot tubes, using clear pvc tubing. Fix the "pitot" ends of the tubes in locations where you think you might like to place your air scoop. Fix the manometer ends up on the handlebars somewhere, all bunched together so you can compare the water level changes directly. That way you get to visually see relative differences at the same moment in time (you don't have to deal with variations in conditions between different runs), and it's cheap as. Like this x5 at once:
DSC_0003.JPG
However, as has been mentioned, the best way (and only sure way to confirm any cooling mod is working), is to measure the actual temperature of your motor. And if you're looking to push performance and run hotter, you should be monitoring motor temperatures anyway.
 
Oh, I never meant to say frying your motor is not fun, its a blast!

I just used to run 3000w all winter, then when the motor would blow its halls, fix it and reduce to 1500w for the summer.

But getting a big fat motor works so much better. Its definitely working better than giving motors so much they cannot use it at all.

DO give your motor as much as it can handle. That's around 2000w for most 500w rated motors. Its just dumb to give too small motors 4000w, as I showed that day at the death race.

It won't measure the internal temp, but you can put a ten buck remote thermometer sensor on the stub of the hub, and it will show you how quick your hub is heating up, and it will show if it heats to equilibrium where heat is leaving as fast as you make it, but you are below that melting point.

This is really all you need to know when pushing the motor hard. If the temp stops going up, you are not going to melt. If it goes up like a rocket, you better stop in 5 min.
 
One other solution to determine the air flow could be the hot-wire anemometer:
A thin wire, made from platinum or nickel or something else is heated up with constant current, and the voltage drop is a funktion of the temperature. And, in our case, the temperature is dependig on the air flow.

This system has improved accuracy at low velocities, compared with pitot.

Use the principle "4 wire measurement" to get improved results.

I made a very rudimentary trial right now, with a "HERAEUS 1 Pt 100 MR 828", supplied with 30 mA constant current.
https://www.reichelt.de/index.html?ACTION=7&LA=3&OPEN=0&INDEX=0&FILENAME=B400%2FMR828_EN.pdf

This prinziple is working fundamentally, but don't complain the sensor if it breaks (it's abused...)
 
Wow, there are really a lot of different ideas/approaches here. Thanks!

The idea is optimizing cooling at higher speeds (about 60mph). The motor is a 3kw hub motor with Statorade and Hubsinks.
Of course it has a temperature sensor and temperatures are monitored. Just measuring motor temperature would not be precise enough to make conclusions if airflow improved or not I think. Conditions would have to be the same and I would need to ride exactly the same manner in both runs.

Ideal would be a wind channel. Unfortunately I don’t have one ;)

Measuring using a pitot tube would affect the results because the tube would change the way the air flows. Not sure if it will affect it significantly.
 
hias9 said:
Wow, there are really a lot of different ideas/approaches here. Thanks!

The idea is optimizing cooling at higher speeds (about 60mph). The motor is a 3kw hub motor with Statorade and Hubsinks.
Of course it has a temperature sensor and temperatures are monitored. Just measuring motor temperature would not be precise enough to make conclusions if airflow improved or not I think. Conditions would have to be the same and I would need to ride exactly the same manner in both runs.

Ideal would be a wind channel. Unfortunately I don’t have one ;)

Measuring using a pitot tube would affect the results because the tube would change the way the air flows. Not sure if it will affect it significantly.


I actually been playing with cooling and prelimiminary tests show drops of 30+ degree C with my dog leash heat sink.... and ATF about instantly... 100*C to 60* C in 10-20 sec. Putting much more power down where I was thermally limited..... Only thing you need is a bottle of water.

Absolutely test the cooling capabilities of a hub motor by looking t the temp data. My sensor is absolutely fast enough, the data is logged, and I have a 100A baseline test data, 100A ATF only test data, and ATF + Heatsink data. All the exact same trip where use 125Wh n the downhill, to town, and 175 Wh on the way back, loaded uphill... It is uncanny how the idential trips can be compared on the log. Identical trips. Time and again, seen t so much I know every throttle blip, every stop sign and every hill I go up through the data. .
 
hias9 said:
Of course it has a temperature sensor and temperatures are monitored. Just measuring motor temperature would not be precise enough to make conclusions if airflow improved or not I think.
Even with just one built in temperature sensor, you can determine if the airflow changes also change the motor temperature in the direction you want, as long as you perform several test runs of each airflow redirection version to average the readings you get.


Now, if you want to determine what specific areas of the motor are changing temperature and to what degree; you would need to instrument the entire motor as thoroughly as possible with many sensors, and record that data with whatever system you would have used for the airflow recording (which would also require instrumenting the motor with many sensors to actually track the airflow).

Or you can use an FLIR camera mounted where it can "see" the motor, you can record this data (or observe it directly on-screen), and not have to instrument the motor *at all*, yet still see the entire motor. If each side is expected to be different results, then you would need two FLIRs to simultaneously meausure both.



So, what data collection system were you going to use to log the airflow measurements?

Almost certainly you can use the same (or equivalent) system to log the temperature measurements.


Measuring the temperature is a MUCH more direct way to get the results you are after, because you don't actually care what the specific airflow is. You only care what the *temperature* is--if the temperature stays low, then the airflow is doing whatever you wanted it to do. ;)


If you insist on measure the airflow, you are probably going to have to invent something to do it that can mount to the spinning motor casing (assuming an outrunner or hubmotor), and then wirelessly transmit the data to a collection system, to log it for your test runs.

But I really think this is overthinking the problem, and you'd get perfectly usable results by simply measuring the motor temperature in one or a few places, then performing several test runs for each version of airflow redirection, collecitng the temperature data for each, averaging the several runs within a version, and then comparing each of these averages for the different versions, to tell you which version worked better.


It might require an iterative approach to designing the airflow redirections, retesting, redesigning, retesting, but it'll get you where you want to be with technology you already have built in, and/or existing cheap technology (wireless temperature sensors and a receiver that can log all of their data together for each test run).


FWIW, anything you use to directly measure airflow is going to change the airflow, becuase it has to be *in* the airflow.

Just like anything you use to directly measure temperature is going to change the temperature, because it is in contact with the thing being measured.

How much each of those thigns changes the measurement depends on the scale of the property being measured, vs the size/shape of the airflow measurement devices, or the mass of the temperature measuring devices, and how many of each there are, and where they are.
 
Well, I guess you are right. It's much easier to try to make 2 comparable runs while only measuring motor temperature.

Are there any Fluid Dynamics simulations available for larger hub motor bikes?
 
You mean a program made just to do it for bikes? Not that I know of.

You could build the files necessary to do the simulation on existing programs, as long as you have one of those programs and a computer powerful enough to run it.

Is probably tedious and time consuming, and you have no way of knowing that the results will match the real world until you build it for real, and then instrument and test the real thing to see if the results are the same.
 
I mean if somebody made simulations for a rear hub motor ebike using existing programs.
Yes, doing that would be very time consuming I think and probably not worth the effort.
Better just trial and error and test it practically.

Is anybody using sheets to direct air to the hub motor and did you notice an improve in cooling at higher speeds?
 
When you're going really fast, like motorcycle speeds and up, then you might get a fair bit of airflow by redirecting, if the "scoops" or whatever shape they are is large enough, and reach far enough into the airflow around the bike to catch the air.

When you're going slow, like normal bicycle speeds, you would probably have to extend the scoops farther out and larger surface area to get the same kind of cooling.


In either case, the redirectors must stick out far enough to be *in* the airstream around the bike. If they are not, there's no airflow for them to redirect, unless you have fans blowing on them.

Where the air flows may also change with speed depending on the particular bike shape, and rider position (legs may get in the way of the airflow at motor height).



IIRC, John in CR has some posts about experiments with this on his bike(s). Justin_LE may also have some experimental data posted around here.
 
hias9 said:
I mean if somebody made simulations for a rear hub motor ebike using existing programs.

~ and test it practically.

~ and did you notice an improve in cooling at higher speeds?

Here is some good data from four trips for you. First set is on the relative flat, a mirror inbound/outbound trip. See the congruence. The difference here is primarily the average speed divfference that 5-6mph makes. The higher average speed (power vs heat vs ect science up in the science lab) gets me a little more heat created. ( See; Average speed) ( to the corner store and back) (2.25 mile, ~50Wh/mile)

The spike ( in speed, blue trace) on the outbound in speed is a hill that I can go down fast.. but on the inbound trip I paused on the uphill cause there was alot of traffic at the moment.. and wanted them to passme, and I didnt want to barrel up the hill at full power and ruin my log ( trying to average my current draw for the test) ... hence the increase/dip in speed there, in the middle of the first two graphs.. )

outbound.PNG inbound.PNG

... and here are ( two sets of ) 4 drag-race style pulls for my bike as I was testing different cooling regimes: On the flat ( actually 1-2% grade, you see that in the final top speed, one direction was ever so slightly uphill) back and forth with a ten second count between pulls.

Heat soak must also alswys be considered. I pause data collections between variable changing just to assure room temp baseline... as much as I can.

You can see the accuracy of the data. You can get familiar with the data. You can see it for what it is. here is two totally different logs.. but.. some of the numbers are uncannily similar and this eliminates the variables for my study.... This is direct power - heat ration sturdy.. Ie how much power can I put down with my various cooling variables without going over a certain threshold.. deriving the best way to cool reliably and repeatability. (85Wh/mile, almost identical data save the heat ) ( same distance, time, Wh used, ect... ) ( one data set was with hubsink, one was without hubsink... variable).

4pulls1.PNG4pulls2.PNG
 
Back
Top