Grin trip simulator core temp

[drbenjamin]

All

trying to decide between a Mac 12T and a BBSHD for my commuter setup. I tried out the Grin trip simulator on their site (which is very cool) and got a surprising result - see below. A long part of my ride is along a river, and I don't expect to use any power there. I set the speed limit to reflect that - along the straight part the watts used show 0, however the core temperature (red line) keeps rising. It makes the temp spike when I hit a steep hill more alarming since the motor is already 35 deg. Is that how geared hubs behave? Do they heat up from pedaling even w/o power?



Thanks!

 

[amberwolf]

If there's no power being used, then there's no heat generated by current flow in a geared hub iwth a freewheeling clutch internally.

Heat from clutch friction should be minimal, no worse than heat in bearings. There's no eddy currents in it while it isn't being powered, as it doesn't spin the motor at all (unlike a DD hub).+

So I don't know what would be causing the simulator to show heating up, unless there is some "hidden factor" that doesn't show in the simulation or it's controls, or you *are* actualy using motor power, even minimally, durign the section of trip that shows the slow rise.

 

[drbenjamin]

Thanks amberwolf. I'm thinking that maybe the MAC is being modeled as a direct drive here - note that in the 1st mile which is all downhill there's still a temp rise, so I don't think that battery power is accounting for the temp rise. If I raise the speed limit high enough the battery starts to be used during the flat parts, but in the pic it is 0W throughout the middle flat section.

 

[markz]

There is the Initial Temperatures section of the graph, which includes ambient temperature, core and shell temp.
If you did not change any settings for the motor itself (Show more, motor parameters) then it sure is strange how the temp is rising.

 

[drbenjamin]

The temps were all at 21 and I selected MAC 12T from the pulldown menu, didn't change any parameters. Looking at them though, I see that eddy currents are non zero and also hysteresis torque. I would think that in a geared hub (with clutch) these terms should be zero when no current is flowing into the motor. I'm guessing that the tool was built for DD hubs and this is a bug affecting geared hubs.

 

[markz]

Grin Tech is quite smart in how they operate, I doubt they'd put out an inferior product that puts out inaccurate information, but it could very well be a bug.

Have you tried using another geared hub and keeping everything else the same?

 

[justin_le]

Thanks amberwolf. I'm thinking that maybe the MAC is being modeled as a direct drive here -

Correct conclusion! The trip simulator doesn't currently have any facility for modeling freewheeling motors, it treats it as though the motor is always engaged since that's a much more interesting scenario to model and explore. If a motor is just freewheeling and not spinning then there is nothing really to model

It IS the case that if you run a MAC or other geared motor even at minimal output torque but at decent RPM's it will still heat up quite a bit just from core losses and that is the source of heat here.

When I get to the point of writing a detailed FAQ on how to use the trip simulator web app it will explain these details. The point of the application isn't really to model a long distance ebike trip and predict how everything will behave in a real world ride. The intention of it is to be a learning tool to understand the ramifications of different usage patterns and motor/controller/battery options in very specific scenarios. If you want to understand the behavior just in the hill climb from km 15 to 18 on your elevation map, then use only that section for modeling, rather than the entire flat section before where you aren't using the motor.

 

[justin_le]

along the straight part the watts used show 0, however the core temperature (red line) keeps rising.

Notice though that even though the electrical power is 0 watts, the motor mechanical power is showing -23W? This is the amount of core losses going into the hub motor to keep it spinning (assuming that it wasn't freewheeling) and that is the exact source of the heat in this web app.

It makes the temp spike when I hit a steep hill more alarming

The spike in temperature on the steep part of the hill when your speed drops to ~12 kph looks to be less than a 20 degree of net temperature rise, which is not at all alarming. Even if the motor core was 100oC at the start of this hill you're still a long ways off from the kinds of temperatures that can cause damage inside the hub (usually >180oC), just to give a bit of perspective.

 

[wturber]

The point of the application isn't really to model a long distance ebike trip and predict how everything will behave in a real world ride. The intention of it is to be a learning tool to understand the ramifications of different usage patterns and motor/controller/battery options in very specific scenarios. If you want to understand the behavior just in the hill climb from km 15 to 18 on your elevation map, then use only that section for modeling, rather than the entire flat section before where you aren't using the motor.

Well it is called a "trip simulator" after all. :^) Maybe an option for geared hubs and mid drives that decouple could be added by the time it leaves beta?

 

[drbenjamin]

The point of the application isn't really to model a long distance ebike trip and predict how everything will behave in a real world ride. The intention of it is to be a learning tool to understand the ramifications of different usage patterns and motor/controller/battery options in very specific scenarios. If you want to understand the behavior just in the hill climb from km 15 to 18 on your elevation map, then use only that section for modeling, rather than the entire flat section before where you aren't using the motor.

Thanks for the reply Justin! And thanks for the fantastic tools you've made for the community, they really helped me understand the parameters that go into selecting ebike components. This is my first build, I chose a geared hub and was concerned about cumulative heating since they shed heat poorly. I segmented the ride as you suggested, and I'm 1/2 way through my build so I'll see soon how things go on our hills! Feeling pretty confident it will all go well.

BTW, am really psyched to see the mid-drives added to the motor simulator. Will they migrate to this tool as well?

 

[justin_le]

Well it is called a "trip simulator" after all. :^)

I know, one of my big regrets! The current name is indeed misleading for what my intentions are. I'm thinking to possibly rename it as "trip modeler" to avoid this. "Time domain EV simulator" doesn't have quite the right ring but would be more accurate. The actual origins of this project was to be just a motor thermal simulator and just show motor heating over time at a given load, but as we kept flushing things out it grew and grew in scope to what we have now.

Maybe an option for geared hubs and mid drives that decouple could be added by the time it leaves beta?

We'll think about it. Part of the problem is that I haven't done testing of motor conductivity parameters for the situation where the motor is static and only the shell is spinning, and I'm not sure how well the current model extrapolates all the way over to the 0rpm case.

We did make one significant update just recently for those who are using the google maps input option by adding an elevation filter slider so that you can smooth out irregularities in the elevation data coming from the google maps API. We saw many scenarios where google's elevation info would result in short sections of like 20% grade, and the simulator model would naturally have the motor stalling out here. It would have the bike climbing at just 1-2 kph (mostly as a result of the human power input) and the motor temperature would skyrocket over a short segment.



So now there is a slider bar that you can use to low pass filter the elevation profile to whatever level of smoothness you want.



And by putting say a 10% smoothing on the raw data you still capture the elevation profile quite well without these glitches appearing which were previously causing the motor to stall. This is the exact same trip as above with the 10% filter active and the results are a lot more realistic: