Erg wattage vs Cycling wattage

[jamesg]

Rowing is not designed for efficiency, being intermittent; as a sport it seems to be somewhere between cycling and lifting.

As far as I know, weight lifting involves lifting something very heavy, plus body weight, a limited number of times; rowing a medium load (100kg?) 200-250 times in a 2km race, and cycling a small load but thousands of times even in a short TT.

If they did not differ, why would anyone want to do all three?

[Nomath]

No point taking the average person on the bike and the average person on the rower you need to take the SAME person on the rower as the bike and just look at HR for a given wattage.

I don't agree. Science is about groups of people, not about individuals. What's the value of finding 30% power difference for a person A, if the result wouldn't apply for B, C and D ?

It should also be verified that the measuring tools are accurate. What do you know about the accuracy of the power measurement in cycling if it isn't on a BikeErg ?

You can criticise my comparison for not having verified that the group of BikeErg users is comparable to the RowErg users. For example, if the BikeErg users are generally older than the RowErg users. So here is the same comparison for age group 40-49. It confirms that the difference is in the order of 10-20%, except for the sprint distances



Another criticism could be that the RowErg users are physically more fit than those using a BikeErg. It's a valid point! The RowErg is certainly used by the rowing elite, whereas I doubt that the cycling elite uses a BikeErg. They probably use more fancy equipment. That's why I hesitate to compare world records or results at the 90th percentile level. However, this elite bias has probably disappeared when we compare at the 50th percentile level, as in the above table.

[MudSweatAndYears]

I think it is not unreasonable to compare mean or median results for a complete group. What is puzzling though, is that for efforts with a duration of 1:00 minute you find that the BikeErg wattages outperform the RowErg performances, while for slightly longer durations (1:40-1:50) it is the other way around. And then, for even longer durations (in excess of 7:00 minutes), the BikeErg again comes out on top.

[MudSweatAndYears]

Another criticism could be that the RowErg users are physically more fit than those using a BikeErg. It's a valid point! The RowErg is certainly used by the rowing elite, whereas I doubt that the cycling elite uses a BikeErg. They probably use more fancy equipment. That's why I hesitate to compare world records or results at the 90th percentile level. However, this elite bias has probably disappeared when we compare at the 50th percentile level, as in the above table.

I would not be too worried about this. Even the rowing elite achieve higher wattages on the BikeErg than on the RowErg (at least for endurance events).
Case in point: Hamish Bond, 30 minutes RowErg = 405 W, 30 minutes BikeErg = 470 W.

[btlifter]

I think it is not unreasonable to compare mean or median results for a complete group. What is puzzling though, is that for efforts with a duration of 1:00 minute you find that the BikeErg wattages outperform the RowErg performances, while for slightly longer durations (1:40-1:50) it is the other way around. And then, for even longer durations (in excess of 7:00 minutes), the BikeErg again comes out on top.

I agree that there seems to be a discrepancy with those times. My gut reaction is that it's a sample-size issue.

My own N1 sample is that the 2k/4k is my crossover point. Shorter than that is faster on the rower (the shorter the distance the more pronounced the difference). Longer is faster on the bike (predictably, the longer the duration the more pronounced the difference).

[Tsnor]

What's the value of finding 30% power difference for a person A, if the result wouldn't apply for B, C and D ?

If the goal is "Carl is happy rowing vs cyclists on zwift" and 30% gets him there, then that's the value from the 30% number.

It's not zwift setting the 30%. It's Carl setting 30%. A cyclist who had bad rowing form might set 40%. A rower who was a crappy rider might set 20% to get paritiy between his rowing and riding.

Good of Carl to share his number so others have a starting point.

Not sure I'd end up using 30%. From the discussion of short vs long the 30% fudge factor might even need to change with distance. Also not sure how zwift users might feel if they learned that other zwift riders ahead of them were not just fudging their weight, but also scaling the measured power. That would keep me on the artificially low end of the percent adjustment scale. (Imagine getting on the top of a leaderboard after you scaled power.. or grabbing a KOM title ?).

[Carl Watts]

30% wouldn't apply to everyone its just a generic starting point.

Heavier rowers are penalized on the Static Erg, there is an increase in lost power just moving the bodyweight, this is not the case on the bike.

Its really not hard to see why there is a big power difference.

1. Your moving your entire bodyweight almost continuously on the rower. This is 100% power loss.

2. Your legs do not assist one another.

3. The power input from your legs is shorter duration and hence peak power required to get that average power is higher on the rower. Cycling is almost a continuous power delivery, rowing is very intermittent and at low ratings its truly horrible on the heartrate for the same pace,

4. Your upper body has to not only try and add power, it has to transfer the power from your legs to the handle.

The end result is a much better all body workout using the rower compared to a bike but its way down in efficiency in transferring power. Should be pretty obvious for anyone who thinks about it in detail.

[Carl Watts]

What's the value of finding 30% power difference for a person A, if the result wouldn't apply for B, C and D ?

If the goal is "Carl is happy rowing vs cyclists on zwift" and 30% gets him there, then that's the value from the 30% number.

It's not zwift setting the 30%. It's Carl setting 30%. A cyclist who had bad rowing form might set 40%. A rower who was a crappy rider might set 20% to get paritiy between his rowing and riding.

Good of Carl to share his number so others have a starting point.

Not sure I'd end up using 30%. From the discussion of short vs long the 30% fudge factor might even need to change with distance. Also not sure how zwift users might feel if they learned that other zwift riders ahead of them were not just fudging their weight, but also scaling the measured power. That would keep me on the artificially low end of the percent adjustment scale. (Imagine getting on the top of a leaderboard after you scaled power.. or grabbing a KOM title ?).

Its not my fudge factor, its a factor that was arrived at by the guy that created the RowedBiker software from his experience both rowing and cycling.

I pretty quickly found out that in reality its not even a fixed percentage and not a linear function, you need a higher correction factor for sprints.

RowedBiker allows for 30%, 40% and 50% correction.

The max I can put out on the rower is 550W for 1 minute and your dead, this sort of power gets you nowhere on a bike sprint without correction.

To many variables to consider, like I said you would need both bits of gear and do your own testing. Some sort of correction curve is required and its gets to complicated. basically Zwift is just for the graphics and a bit f a change from RowPro, your not interested in gaining titles in the D Cat.

[Tsnor]

My own N1 sample is that the 2k/4k is my crossover point. Shorter than that is faster on the rower (the shorter the distance the more pronounced the difference). Longer is faster on the bike (predictably, the longer the duration the more pronounced the difference).

Are you measuring power on an indoor trainer or through outdoor power meter ? If indoors the 2k/4k crossover might be an artifact of underreported cycling power/inability to get full power on a fixed vertical cycle.

This video demonstrates then tries to explain a 200W or more difference on trainer vs outdoor power meter for sprints while showing that they match perfectly for 5 minute + durations. His sprint at 1300W to 1400W would be in your range. https://www.youtube.com/watch?v=HGvpLzC2el4

[btlifter]

My own N1 sample is that the 2k/4k is my crossover point. Shorter than that is faster on the rower (the shorter the distance the more pronounced the difference). Longer is faster on the bike (predictably, the longer the duration the more pronounced the difference).

Are you measuring power on an indoor trainer or through outdoor power meter ? If indoors the 2k/4k crossover might be an artifact of underreported cycling power/inability to get full power on a fixed vertical cycle.

This video demonstrates then tries to explain a 200W or more difference on trainer vs outdoor power meter for sprints while showing that they match perfectly for 5 minute + durations. His sprint at 1300W to 1400W would be in your range. https://www.youtube.com/watch?v=HGvpLzC2el4

I only measure power from the bike erg. And actually, you raise an important point: I have been told by several persons that the bike erg under-reports power at high wattages. I have no explanation for that though.

[MartinSH4321]

I think it is not unreasonable to compare mean or median results for a complete group. What is puzzling though, is that for efforts with a duration of 1:00 minute you find that the BikeErg wattages outperform the RowErg performances, while for slightly longer durations (1:40-1:50) it is the other way around. And then, for even longer durations (in excess of 7:00 minutes), the BikeErg again comes out on top.

I agree that there seems to be a discrepancy with those times. My gut reaction is that it's a sample-size issue.

My own N1 sample is that the 2k/4k is my crossover point. Shorter than that is faster on the rower (the shorter the distance the more pronounced the difference). Longer is faster on the bike (predictably, the longer the duration the more pronounced the difference).

Have you ever tested 200m and max watts on the bike erg? If I remember right the WR for max watts on the bike is almost 2400W on a bike (Manfred Nürscheler) and on the static rower 0:59 pace or around 1700W, so almost 50% more on the bike but only for a few seconds.
This sounds right for me as the bike movement is more efficient and legs can produce a lot of force when anaerobic alactic, but lactic acid soon slows you down while on the rower you also have back and arm muscles to work with, so faster rows for 1'-4'.

[btlifter]

I think it is not unreasonable to compare mean or median results for a complete group. What is puzzling though, is that for efforts with a duration of 1:00 minute you find that the BikeErg wattages outperform the RowErg performances, while for slightly longer durations (1:40-1:50) it is the other way around. And then, for even longer durations (in excess of 7:00 minutes), the BikeErg again comes out on top.

I agree that there seems to be a discrepancy with those times. My gut reaction is that it's a sample-size issue.

My own N1 sample is that the 2k/4k is my crossover point. Shorter than that is faster on the rower (the shorter the distance the more pronounced the difference). Longer is faster on the bike (predictably, the longer the duration the more pronounced the difference).

Have you ever tested 200m and max watts on the bike erg? If I remember right the WR for max watts on the bike is almost 2400W on a bike (Manfred Nürscheler) and on the static rower 0:59 pace or around 1700W, so almost 50% more on the bike but only for a few seconds.
This sounds right for me as the bike movement is more efficient and legs can produce a lot of force when anaerobic alactic, but lactic acid soon slows you down while on the rower you also have back and arm muscles to work with, so faster rows for 1'-4'.

Hahaha. Yes, and my max wattage is about 15% higher on the rower than the bike erg.

But, again, I think this is a "bike erg" specific issue. Though I have no personal experience, from speaking to others my max wattage would be much high (25ish%, maybe) on a Watt bike or other (presumably) a real bicycle.

***This discrepancy between the erg and other bike trainers seems to disappear for efforts longer than about 1 minute***

[JaapvanE]

No point taking the average person on the bike and the average person on the rower you need to take the SAME person on the rower as the bike and just look at HR for a given wattage.

I don't agree. Science is about groups of people, not about individuals. What's the value of finding 30% power difference for a person A, if the result wouldn't apply for B, C and D?

It is much more interesting to know that on average a person will do 30% better, with standard deviation X, etc. If you do cohort comparison you run the risk of outliers cancelling out with other outliers (as you are comparing the mean of the group on a rowerg with the mean on a bike). If you calculate from the individual first and generalize to the group later you retain these outliers.

[Carl Watts]

Have you ever tested 200m and max watts on the bike erg? If I remember right the WR for max watts on the bike is almost 2400W on a bike (Manfred Nürscheler) and on the static rower 0:59 pace or around 1700W, so almost 50% more on the bike but only for a few seconds.
This sounds right for me as the bike movement is more efficient and legs can produce a lot of force when anaerobic alactic, but lactic acid soon slows you down while on the rower you also have back and arm muscles to work with, so faster rows for 1'-4'.

My observations even after only a session or two on Zwift using the rower was that a correction curve was required not a fixed percentage.

Losses on the rower decreased at slower paces like 2:30 to like 15% and at around the 2:00 pace you need 30% to all out sprints you needed a 50% boost

The other problem on the Erg is you have no gears and you cannot even change the drag factor during the row so trying to sprint using your steady state rowing drag factor which is always the case as bike rides are typically quite long is a problem when it comes to sudden sprints.

The power conversion math is not simple and the only way to do it really would require you to do individual testing on the bike and the rower and put at least 3 or 4 test points into forming a correction graph. Ultimately this is just all to hard and probably the reason that Zwift officially didn't do it and left it to 3rd parties to just derive something simple that still allows you to join in and row on Zwift and just enjoy the graphics while not taking it to seriously.

[Nomath]

It is much more interesting to know that on average a person will do 30% better, with standard deviation X, etc. If you do cohort comparison you run the risk of outliers cancelling out with other outliers (as you are comparing the mean of the group on a rowerg with the mean on a bike). If you calculate from the individual first and generalize to the group later you retain these outliers.

You are right in theory, but only in theory!
For your approach you have to scan the individuals who listed their seasonal PR's both in the RowErg rankings and in the BikeErg rankings, The RowErg rankings consists of roughly 16,000 individuals ; the BikeErg rankings of about 1000 individuals. If you find the same person in both rankings you might find comparable times to calculate the power discrepancy. I don't have the software to do this search, nor the ambition to do this tedious work. It's a huge task. I believe my comparison-of-medians approach provides a credible estimate of the power differences. It's far below the conjectured 30%.

Please go ahead if you see a hidden treasure trove in the ranking data.