Methodology proposal to establish the assistance levels in an E Bike

[AntonioAlfaro]

Hello, allow me to share with you a first draft document that I have prepared to guide in the determination of the levels of assistance in an EBike. It is not a final document, as soon as I receive my Domane+ SLR6 I will finish field checking and correct it as necessary.
If you cycle consistently and also use the assistance of the E Bike as a complement to your personal strength, you will surely want to use just enough of your EBike battery. This will allow you to go further on your rides or take more demanding rides.
I hope it's of your interest

 

[Stefan Mikes]

Antonio,

Unfortunately, it is hard for me to accept your approach of W/kg as the universal backbone of your theory.

A word about myself first:
I am 62 years of age who annually rides for 8000-10000 km on e-bikes. My multiple medical condition combined with medications limit my FTP to, say, 93 W. As I weigh 96 kg, my coefficient would be 0.97 W/kg. Dr Collang seems to have never done a study of a category below "Untrained"; that category should be named "A senior E-biker". Otherwise, you might think I were some "Subhuman"? If I could provide as many W/kg as Dr. Collang assigned to the "Untrained Male" then sorry but I would have never needed an e-bike!

I my opinion, the W/kg approach makes little sense for e-bikes, and typical users of these. Far better approach -- in my opinion -- would be:

• Determining the necessary total energy input into the ride (distance, elevation gain, total system weight, average speed deserved, wind conditions, frequency of starts/stops, etc)
• Subtracting the rider's power input from the above
• Now, you could calculate the necessary e-bike assistance required.

For rides in the flat area, the parameter of W/km would be far more practical. If the ride were on a loop route in the mountains, the climbs would be partly offset by descents (provided the e-biker would not pedal on descents, and would avoiding braking downhill -- but the latter could not be assumed for most of riders).

Let me give two examples based on two different rides last weekend with the data measured by two different Specialized e-bikes and stored in a Wahoo bike GPS computer and interpreted by Wahoo Companion App.

Ride A
(A Workout on a Vado SL, a low power fitness e-bike)

• Motor: 35 Nm max, 240 W Max Motor Power (mechanical), Speed Restriction: 25 km/h
• Distance ridden: 100.12 km
• Elevation gain: 183 m
• Wind: The effect cancelled by riding a loop (steady wind from one direction)
• Average net speed: 20.6 km/h
• Rider's Normalized Power (measured): 93 W
• Rider's Max Power (measured): 396 W
• Rider's energy input: 1318 kJ (366 Wh)
• Battery input (electrical): 321 Wh (measured)

To keep it simple, the rider's contribution into the ride was 100 * (366 / (366 + 321)) = 53.27%
The battery consumption factor: 3.21 Wh/km

Ride B
(A Workout with Road Cyclists and then slower returning home on a Vado 6.0, a 45 km/h S-Pedelec)

• Motor: 90 Nm max, 520 W Max Motor Power (mechanical), Speed Restriction: 45 km/h
• Distance ridden: 105.08 km (of which 58 km was the Workout)
• Elevation gain: 229 m
• Wind: mild from constant direction
• Average speed: 24.6 km/h; average speed on the 58 km workout part: 27.1 km/h; max speed on the flat: 47 km/h (sprint).
• Rider's Normalized Power (measured): 84 W
• Rider's Max Power (measured): 475 W
• Rider's energy input: 908 kJ (252 Wh)
• Battery input (electrical): 658 Wh (measured)

The rider's contribution into the ride was 100 * (252 / (252 + 658)) = 27.69%
The battery consumption factor: 6.26 Wh/km.

Could you please @Antonio interpret the data as above according to your theory?

 

[AntonioAlfaro]

Thank you Stefan for your feedback.
First of all I must say that a person's FTP value is nothing more than an indication of their threshold level of power that they are able to sustain during one hour (just a metric). A low or high value is simply an indication of what we might expect in terms of cycling performance. Let's look at the extremes: a Grand Tour cyclist could tackle long distance, steep slopes at fairly high speeds. On the other extreme, a cyclist with a very, very low FTP can do something similar, but getting almost 100% of the energy needed from his EBike's battery. This last one is not bad; it is simply a scenario where the EBike is working almost as Class 2 electric bike.
As far as the cyclist supplies more energy (higher FTP), the energy supplied by the Ebike will be less.
If your W/Kg ratio is 0.97, it does not mean anything other than that you are able to do some cycling by yourself, but if the route becomes somewhat demanding, the EBike's battery will provide a lot of energy during the tour. If your tours are high mountains, the battery will last very little.
The wonderful thing about Ebikes is exactly that, allowing all kinds of people to enjoy cycling by supplying the lack of energy that the cyclist may have. Otherwise we would be very limited or unable to do recreational cycling at all. Of course young, strong and healthy people do not require Ebikes yet, but there is a high population of cycling enthusiasts who will thank God for having an eBike.
Looking at the data you provided for Ride A and B, it's clear that both rides were on flat terrain and probably some downhill (don't know), I see considerable battery power consumption which is consistent with the low FTP. In general, all in line with laws of physics governing cycling.
Using 80% of your FTP, your mass, Domane+ mass, 2 extra Kg, flat terrain, you are able to ride at 20Km/h by your own power.

Finally let me use your FTP and Mass to simulate on a Domane+ SLR, I would get the following results:
FTP= 93W
Cyclist mass = 93Kg
Total mass = 110Kg
Equalized Power = 386W
Assistance Max Power = 311W
ECO level = 150W
MID level = 225W
MAX level = 300W
If you look, these results are very close to the default settings that TQ proposes

 

[Stefan Mikes]

Antonio, so please have a look to the most intensive mountain ride I had on the 45 km/h older Vado 5.0:

The Most Remote Southeast Road Point of Poland | Strava

BLEvo v3.9.6 Android - Battery consumed: 174% (981Wh) - Average consumption per KM: 7.85 Wh/km - Percentage of average assistance: 49.1% - Wh ride: 1322.1Wh - Biker: 25.8% (341.1Wh) - Battery: 74.2% (981Wh) | Strava

www.strava.com

You will find all the measured parameters there. Interestingly, I provided 25.8% of own contribution, and neither of my e-bikes has a throttle, so the comparison to a Class 2 e-bike is not really honest!

What about this?

My First Double Metric Century! - Stefan M's 208.6 km bike ride

That was A ride!

www.strava.com

Not sure how much energy I drew from the batteries but one thing is clear: I was assisted by a strong tailwind on that ride!

 

[rdv]

Thank you Stefan for your feedback.
First of all I must say that a person's FTP value is nothing more than an indication of their threshold level of power that they are able to sustain during one hour (just a metric). A low or high value is simply an indication of what we might expect in terms of cycling performance. Let's look at the extremes: a Grand Tour cyclist could tackle long distance, steep slopes at fairly high speeds. On the other extreme, a cyclist with a very, very low FTP can do something similar, but getting almost 100% of the energy needed from his EBike's battery. This last one is not bad; it is simply a scenario where the EBike is working almost as Class 2 electric bike.
As far as the cyclist supplies more energy (higher FTP), the energy supplied by the Ebike will be less.
If your W/Kg ratio is 0.97, it does not mean anything other than that you are able to do some cycling by yourself, but if the route becomes somewhat demanding, the EBike's battery will provide a lot of energy during the tour. If your tours are high mountains, the battery will last very little.
The wonderful thing about Ebikes is exactly that, allowing all kinds of people to enjoy cycling by supplying the lack of energy that the cyclist may have. Otherwise we would be very limited or unable to do recreational cycling at all. Of course young, strong and healthy people do not require Ebikes yet, but there is a high population of cycling enthusiasts who will thank God for having an eBike.
Looking at the data you provided for Ride A and B, it's clear that both rides were on flat terrain and probably some downhill (don't know), I see considerable battery power consumption which is consistent with the low FTP. In general, all in line with laws of physics governing cycling.
Using 80% of your FTP, your mass, Domane+ mass, 2 extra Kg, flat terrain, you are able to ride at 20Km/h by your own power.

View attachment 154783

Finally let me use your FTP and Mass to simulate on a Domane+ SLR, I would get the following results:
FTP= 93W
Cyclist mass = 93Kg
Total mass = 110Kg
Equalized Power = 386W
Assistance Max Power = 311W
ECO level = 150W
MID level = 225W
MAX level = 300W
If you look, these results are very close to the default settings that TQ proposes

View attachment 154787

These maxima, whether 150/225/300W, or 154/249/300W paired to 98%, 137%, and 200% seems preferable to the draft's 74/102/142W. They would suit a 150W rider well, provide headroom for when the 104W rider is able to contribute some additional bursts of power.

 

[AntonioAlfaro]

These maxima, whether 150/225/300W, or 154/249/300W paired to 98%, 137%, and 200% seems preferable to the draft's 74/102/142W. They would suit a 150W rider well, provide headroom for when the 104W rider is able to contribute some additional bursts of power.

Very good. The idea of optimizing support levels is to get the most out of every watt of the battery so that we always feel like we are doing enough of our part and that support is for those moments of insufficiency. In this way the battery will last us for more travel.
I would like to know if you can do a calculation with my proposal using your personal mass and FTP values, and compare it with your current level settings. It would be an interesting exercise.

 

[rdv]

Not sure how to go about the calculation, but happy to pass on my data. I don't have a proper FTP, but I'd say it's 100W by default, in that Angina doesn't let me sustain anything higher. At age 75, it's unlikely to get better. My combined rider and D+6 and accessories is 90 kg.
I'm down to small tweaks. My latest settings are: E 125W/100%, M 225W/175%, H 300W/200%. The only way to tune at this point is to see how they feel on the road, which I will do this weekend.

 

[AntonioAlfaro]

Not sure how to go about the calculation, but happy to pass on my data. I don't have a proper FTP, but I'd say it's 100W by default, in that Angina doesn't let me sustain anything higher. At age 75, it's unlikely to get better. My combined rider and D+6 and accessories is 90 kg.
I'm down to small tweaks. My latest settings are: E 125W/100%, M 225W/175%, H 300W/200%. The only way to tune at this point is to see how they feel on the road, which I will do this weekend.

Using my proposed methodology and your data:
FTP = 100W
Total Mass=90Kg

we got Equalized Power = Total Massx3.5= 90x3.5=315W,
and Assistance Max Power=(Equalized Power – FTPX80%)=(315-100x.8)=235W

ECO= 235x50%=125W
MID= 235X75%=225W
HIGH=235X100%=235W

These are the starting optimal assistance levels for your data. It would be very valuable if you can have a test and send feedback. Thanks a lot.

 

[mschwett]

I think this is somewhat interesting, but seems flawed to me first in the reliance on mass as a primary driver, and secondly the abstraction of the ”slope” of the ride (really the total amount of climbing per distance is the factor, as most rides start and end in the same place) . The w/kg metric really only matters when one is riding uphill, as your methodology acknowledges with the “no assistance” on flat criteria. On flat ground, desired speed vs drag is really the criteria, and most recreational e-bikers choose to burn up a lot of juice to go a little further, a little faster. Here, drag is the enemy and your calculation in Stefan’s case from BikeCalculator assumes a far lower friction and drag profile than a recreational rider on a vado presents.

I measure power for all my rides, whether on my creo or aethos (a non-e bike), and my goal is typically to maintain a consistent power output, near my tempo level, for the entire ride. This level is limited greatly by a heart rhythm condition and the attendant medications, but other than that I’m in good physical condition. My FTP is only around 230w, but i can sustain an actual average of 200w (not weighted or normalized) for many hours. By minimizing drag and rolling resistance, I routinely ride 80 to 140 km at an average power output of 200w at an average speed of between 24-29 kph for rides with 40-80 feet of climbing per mile (Sorry for the imperial units there.) From this one could say that a similar rider outputting only half as much - 100w - would only need 100w of effective motor assistance (120w at the battery) for that same period, resulting in a battery consumption of only around 4wh/km, or 1/3 lower than Stefan’s ride “b” consumption despite several times as much climbing and a much higher average speed! My point in this is to simply say that w/Kg is really not the driver. Losses to drag are, any they correlate very directly to the bike, the rider, and the speed. Any methodology needs to take those things into account first.

The reason Coggan and others were so deliberate about w/Kg is that in the enthusiastic road rider or professional road rider world, drag has already been minimized with low riding positions, smooth and tight fitting clothing and helmets, deep profile wheels, relatively narrow tires, and aerodynamically optimized frames. Within that context, and the climbing of a serious race, w/Kg makes sense. For a recreational E-biker, I do not think it‘s meaningful. I’d suggest a point based system. Add points for desired speed, exponentially. MultipLy by points for drag, including riding surface and rider position. Translate points to w/km?

 

[Jeremy McCreary]

Honestly, I don't know how you can ignore aerodynamic power losses and expect to come up with a realistic scheme for tuning ebike assistance levels for maximum battery range.

The power lost to air resistance is seldom negligible in cycling, and it's the main power loss in many common riding scenarios — even for riders in racing garb and position with no headwind.

Since aerodynamic losses depend very strongly on ground speed and headwind speed but only very weakly on bike and rider weight (mainly via bike and rider frontal areas), they're largely uncaptured in Coggan's rather unusual "W/kg" approach.

I'd urge you to study Wilson and Schmidt, 2020, Bicycling Science, 4th edition very carefully before proceeding. The chapters on human power generation and power losses in cycling will be especially helpful. These apply to any bike, electric or otherwise.

Using a spreadsheet and the formulas and empirical data in Wilson and Schmidt, you can estimate total power loss on hard pavement based on ground speed, headwind, rider position, rider clothing, road gradient, and tire type. Then you can subtract out the rider's estimated power contribution at the crank to get the power shortfall (if any) the motor will have to cover.

Only then will you be in a position to optimize assist levels for battery range in a specific riding scenario — say, 5% grade in a 5 m/s headwind in upright position in street clothing. I think you'll find that there are too many free parameters in play to come up with a single optimization for general riding — especially on a bike with only 3 assist levels to work with.

This post might also be useful. Good luck.

 

[AntonioAlfaro]

Hello everyone,
I really appreciate your comments. Some of them express a concern that my proposal is not considering all the parameters that influence cycling. I must tell you that they are directly considered when limiting the speed at which you can move with each assistance levels. The determination of the assistance levels of my proposal is based on FTP, Total Mass, slope of the route and speed of travel ( a reasonable speed for pedal-assisted cycling) .
If for example we take the simulation I did for "rdv"
FTP = 100W
Total Mass=90Kg

we got Equalized Power = Total Massx3.5= 90x3.5=315W,
and Assistance Max Power=(Equalized Power – FTPX80%)=(315-100x.8)=235W

ECO= 235x50%=125W
MID= 235X75%=225W
HIGH=235X100%=235W

You can verify that for the range of slope from 0% to 3% in which assistance is not required, "rdv" could move between 22 Km/h max and 8.8 Km/h min, this due to the effect of all those other forces and parameters that you mention. So for the rest of levels and slopes:

ECO:
3% slope 19Km/h ...........; ... .... 7% slope 10Km/h
MID:
7% slope 13Km/h ...........; .........11% slope 8.8Km/h
HIGH:
11% slope 10.8Km/h ........, ........20% slope 6.2 Km/h

I am only attaching an image, but you can do any verification.

 

[AntonioAlfaro]

Hello everybody,
There is a concern about the discrepancy in the power level reading on the TQ system display, attached response received from TQ support.

 

[Calcoaster]

Hello everybody,
There is a concern about the discrepancy in the power level reading on the TQ system display, attached response received from TQ support.

View attachment 154866

Sounds like a reply from marketing, not tech. I understand that it could be tricky to extrapolate an accurate power value from their torque sensor, but cadence should be straight forward enough since their motor system must use a cadence sensor.

 

[rdv]

Hello everyone,
I really appreciate your comments. Some of them express a concern that my proposal is not considering all the parameters that influence cycling. I must tell you that they are directly considered when limiting the speed at which you can move with each assistance levels. The determination of the assistance levels of my proposal is based on FTP, Total Mass, slope of the route and speed of travel ( a reasonable speed for pedal-assisted cycling) .
If for example we take the simulation I did for "rdv"
FTP = 100W
Total Mass=90Kg

we got Equalized Power = Total Massx3.5= 90x3.5=315W,
and Assistance Max Power=(Equalized Power – FTPX80%)=(315-100x.8)=235W

ECO= 235x50%=125W
MID= 235X75%=225W
HIGH=235X100%=235W

You can verify that for the range of slope from 0% to 3% in which assistance is not required, "rdv" could move between 22 Km/h max and 8.8 Km/h min, this due to the effect of all those other forces and parameters that you mention. So for the rest of levels and slopes:

ECO:
3% slope 19Km/h ...........; ... .... 7% slope 10Km/h
MID:
7% slope 13Km/h ...........; .........11% slope 8.8Km/h
HIGH:
11% slope 10.8Km/h ........, ........20% slope 6.2 Km/h

I am only attaching an image, but you can do any verification.

View attachment 154863

Though we all have e-bikes in mind, we're coming at this from different perspectives that, not surprisingly, reflect our own current health, capabilities, priorities and goals. Due to cardio-vascular limitations, mine are to ride and climb without inducing debilitating chest pain. On the other hand, speed is secondary and slow such that aerodynamic drag is of little concern to me, but significant to others. While I'll occasionally enjoy the breezy feel of assist in the flats, my main goal is to again climb the hills I rode up 7-8 years ago on my own, generally up to the 10-12%, rarely reaching 15%. Others here have quite different goals.
Despite Antonio's encouragement, I have no intention of tackling a 20% grade at 6.2 km/h or any other speed. But although I've already honed in on my motor preferred tuning settings, I'll give Antonio's suggestions a sporting try (perhaps this Saturday).

Regarding TQ feedback:
I've reached to TQ a couple of times and have given up, having received the same kind of generic, uninformative Customer Relations responses that add no useful knowledge. What they publish online is no better. If anyone knows how to reach a technical representative that can offer true insights and/or follow up on constructive suggestions, please advise. One aspect that needs attention is the amount of motor assist power reported on the display, unless mine is faulty. It varies wildly several times per second by up to 30-40W, making it practically useless. It desperately needs to be settled down through some sort of 5-10 second averaging. It would also help to have a better sense of how the assist algorithms are structured. Despite all this discussion about max Power assist, I see very few values close to the maximums on the display, and they are fleeting.

Antonio a couple of observations:
MID = 0.75 x 235 ~ 175W or intentionally overridden by 225W (HIGH only 10W higher)?
What % assist should be assigned for each level?
How does the length of the climbs affect your methodology?

 

[AntonioAlfaro]

Though we all have e-bikes in mind, we're coming at this from different perspectives that, not surprisingly, reflect our own current health, capabilities, priorities and goals. Due to cardio-vascular limitations, mine are to ride and climb without inducing debilitating chest pain. On the other hand, speed is secondary and slow such that aerodynamic drag is of little concern to me, but significant to others. While I'll occasionally enjoy the breezy feel of assist in the flats, my main goal is to again climb the hills I rode up 7-8 years ago on my own, generally up to the 10-12%, rarely reaching 15%. Others here have quite different goals.
Despite Antonio's encouragement, I have no intention of tackling a 20% grade at 6.2 km/h or any other speed. But although I've already honed in on my motor preferred tuning settings, I'll give Antonio's suggestions a sporting try (perhaps this Saturday).

Regarding TQ feedback:
I've reached to TQ a couple of times and have given up, having received the same kind of generic, uninformative Customer Relations responses that add no useful knowledge. What they publish online is no better. If anyone knows how to reach a technical representative that can offer true insights and/or follow up on constructive suggestions, please advise. One aspect that needs attention is the amount of motor assist power reported on the display, unless mine is faulty. It varies wildly several times per second by up to 30-40W, making it practically useless. It desperately needs to be settled down through some sort of 5-10 second averaging. It would also help to have a better sense of how the assist algorithms are structured. Despite all this discussion about max Power assist, I see very few values close to the maximums on the display, and they are fleeting.

Antonio a couple of observations:
MID = 0.75 x 235 ~ 175W or intentionally overridden by 225W (HIGH only 10W higher)?
What % assist should be assigned for each level?
How does the length of the climbs affect your methodology?

Hi rdv,
My apologies, there is an error. MID level is 235*75%=175 W. Thanks a lot.
for now I can not suggest the % of assistance. It's something I'll have to try when I get my Domane+. Please use the ones you are used to and let us know.

 

[mschwett]

...

Regarding TQ feedback:
I've reached to TQ a couple of times and have given up, having received the same kind of generic, uninformative Customer Relations responses that add no useful knowledge. What they publish online is no better. If anyone knows how to reach a technical representative that can offer true insights and/or follow up on constructive suggestions, please advise. One aspect that needs attention is the amount of motor assist power reported on the display, unless mine is faulty. It varies wildly several times per second by up to 30-40W, making it practically useless. It desperately needs to be settled down through some sort of 5-10 second averaging. It would also help to have a better sense of how the assist algorithms are structured. Despite all this discussion about max Power assist, I see very few values close to the maximums on the display, and they are fleeting.

...

i've heard quite a few complaints about the power readings of the TQ system. hope they get it sorted.

in the meantime, have you tried pairing via bluetooth or ANT+ to a cycling app or computer that will display smoothed/averaged power? it would be interesting to know if the values are just spiky and basically correct, or ... wrong. even good power meter data is pretty spiky, but what you're describing sounds pretty bad. you should also be able to easily peg the motor output in the higher assist settings, but that value won't be shown anywhere except the top tube display or the TQ app.

 

[AntonioAlfaro]

Hello everyone,
I understand that TREK has an app through which different data could be seen during the tour. What I don't know is if at the end of the journey all that information is loaded or can be loaded to a dashboard where it can be analyzed. Or if you could have access to any JSON file to analyze it externally. Someone knows something about it?

 

[rdv]

i've heard quite a few complaints about the power readings of the TQ system. hope they get it sorted.

in the meantime, have you tried pairing via bluetooth or ANT+ to a cycling app or computer that will display smoothed/averaged power? it would be interesting to know if the values are just spiky and basically correct, or ... wrong. even good power meter data is pretty spiky, but what you're describing sounds pretty bad. you should also be able to easily peg the motor output in the higher assist settings, but that value won't be shown anywhere except the top tube display or the TQ app.

As someone who started his career with analog computer (slide rule), I'll have to defer to Antonio, when he gets his +SLR 6, to dig a little deeper. Another handicap is that my Karoo 2 head unit only records my power input, credibly, but not the motor's contribution. I expect they'll eventually sort that out. In the meantime I'll give the phone Trek app a try. Don't mean to give Antonio false hope, but delivery of my +SLR 6 was promised in 4 months, but arrived after 2.

 

[AntonioAlfaro]

As someone who started his career with analog computer (slide rule), I'll have to defer to Antonio, when he gets his +SLR 6, to dig a little deeper. Another handicap is that my Karoo 2 head unit only records my power input, credibly, but not the motor's contribution. I expect they'll eventually sort that out. In the meantime I'll give the phone Trek app a try. Don't mean to give Antonio false hope, but delivery of my +SLR 6 was promised in 4 months, but arrived after 2.

hello rdv;
I also have a Karoo2 that I bought with great enthusiasm for all the features it offered. But everything changed for me when SRAM acquired Hammerhead and divorced Garmin and Shimano. since then I can't see data from my Ultegra DI2, nor can see data from my Garmin power pedals. I think I have to go back to the Garmin Edge bike computer.

 

[rdv]

hello rdv;
I also have a Karoo2 that I bought with great enthusiasm for all the features it offered. But everything changed for me when SRAM acquired Hammerhead and divorced Garmin and Shimano. since then I can't see data from my Ultegra DI2, nor can see data from my Garmin power pedals. I think I have to go back to the Garmin Edge bike computer.

I was able to sideload the Ki2 workaround (highlighted by DC Rainmaker and Shane Miller) onto the Karoo 2. It shows the front and rear chain positioning very nicely. Not sure of the origin, but I also get a nice two beep alert when the next rear gear would trigger a concurrent Synchro 2 front and rear shift.

Hammerhead a also added an e-bike specific set or parameters, with mixed success so far for the Trek Domane+ SLR6. The K2 reports very nicely on speed and main battery status, but not combined power. I haven't looked tried cadence.

Surprising about the Garmin pedals, I thought the hangup was only with Shimano.