Zwift vs TrainerRoad Virtual Power differences

cgfw201

So I've been Zwifting on the iOS beta for the last 5 days after a few years of TrainerRoad.

Zwift is brilliant, loads of fun and genuinely look forward to jumping on the turbo and want to keep riding at the end of sessions.

Anyway, I've got a non-smart trainer (Elite Fluid Chrono) which I had built my FTP up to 284 on via TR, with most recent FTP test a month or so ago.

On Zwift I find I'm able to hold 350+ watts for way longer than I felt I could on TR. It could be because this is in a race rather than just man vs screen.

This was backed up by a race performance this morning on Zwift where I averaged 380 watts for 13 minutes...
https://www.strava.com/activities/77500 ... 8963603946

Should I trust this data or am I doing something wrong? Power meter on the shopping list for Q1 next year, until then I'm on Zpower.

(This was most TR ftp test https://www.trainerroad.com/cycling/rid ... bber-glove)

fat daddy

disclaimer about zpower on zwift site :-

zPower's wattage estimate is intended as a bench mark for progress. It's not an absolute measurement and will likely differ from your actual power.

I don't think its particularily accurate as zpower is guessing the power based upon what it knows about your exact trainer and the sensors you have attached. The assumes size and pressure of your tires.

zpower is really as they said a benchmark to see your improvement AND to get your bike moving

cgfw201

fat daddy wrote:

disclaimer about zpower on zwift site :-

zPower's wattage estimate is intended as a bench mark for progress. It's not an absolute measurement and will likely differ from your actual power.

I don't think its particularily accurate as zpower is guessing the power based upon what it knows about your exact trainer and the sensors you have attached. The assumes size and pressure of your tires.

zpower is really as they said a benchmark to see your improvement AND to get your bike moving

Yeh understand all that, just wondering why it seems to be so different from TrainerRoad who must be using very similar data to calculate the virtual power from the resistance curve of my trainer?

markhewitt1978

It's all estimates. The Zwift power curve is just from there own experiments with each trainer

cgfw201

markhewitt1978 wrote:

It's all estimates. The Zwift power curve is just from there own experiments with each trainer

cool. will take it. main thing it's consistent to itself. do suspect I'll have a shock when I get a PM and these numbers aren't quite the same in reality.

bianchimoon

I'm by no means an expert but 350 watts for 15 mins would put you in quiet an elite category. I suspect your zwift settings could be wrong i.e. is your trainer listed? did you input correct weight. here's a snip from cyclist mag
"As for road racers, some of the top athletes on the domestic pro circuit can be expected to exceed 380watts for a 20-minute effort at a similar weight to Froome" With your figures from zpower you would be looking at doing a 10mile TT in circa 22 mins? I could probably do it in 30 mins which equates to 161 watts (very similar to my zpower output)

of course you could be a top athlete? regardless it's a great way to train over the winter

cgfw201

bianchimoon wrote:

I'm by no means an expert but 350 watts for 15 mins would put you in quiet an elite category. I suspect your zwift settings could be wrong i.e. is your trainer listed? did you input correct weight. here's a snip from cyclist mag
"As for road racers, some of the top athletes on the domestic pro circuit can be expected to exceed 380watts for a 20-minute effort at a similar weight to Froome" With your figures from zpower you would be looking at doing a 10mile TT in circa 22 mins? I could probably do it in 30 mins which equates to 161 watts (very similar to my zpower output)

of course you could be a top athlete? regardless it's a great way to train over the winter

My weight and trainer are correct.

I've done 10m TTs in 23 mins before (on a road bike) so your numbers above aren't a mile off.

The only thing I'm remotely concerned/surprised by is that my wattage seems to be so much more on Zwift than TR.

bianchimoon

cgfw201 wrote:

bianchimoon wrote:

I'm by no means an expert but 350 watts for 15 mins would put you in quiet an elite category. I suspect your zwift settings could be wrong i.e. is your trainer listed? did you input correct weight. here's a snip from cyclist mag
"As for road racers, some of the top athletes on the domestic pro circuit can be expected to exceed 380watts for a 20-minute effort at a similar weight to Froome" With your figures from zpower you would be looking at doing a 10mile TT in circa 22 mins? I could probably do it in 30 mins which equates to 161 watts (very similar to my zpower output)

of course you could be a top athlete? regardless it's a great way to train over the winter

My weight and trainer are correct.

I've done 10m TTs in 23 mins before (on a road bike) so your numbers above aren't a mile off.

The only thing I'm remotely concerned/surprised by is that my wattage seems to be so much more on Zwift than TR.

have you changed tyre at all since changing progs? if not could be s simple algorithm issue, since it's not actual power just use the figures you feel most comfortable with, it's all about improving them anyway not comparing

okgo

Why don't you buy a or borrow a powermeter, why people pay these figures any attention I have no idea, they're almost always completely fictional, especially when comparing two different models of working!

23 mins on a roadbike isn't a barometer for much, course, position, height, etc all play a huge part.

denkfaul

I've got an Elite Qubo Power Fluid which probably has a similar resistance curve to yours, and definitely noticed a difference in virtual power being reported by TR and Zwift. The range between 150 - 300W was reasonably similar, but anything outside this range was pretty different. Interestingly enough now that I have a real power meter I've done some quick tests, and again between the 150 - 300W I found only 10-15W difference between virtual and real, but again outside of this range it varied a reasonable amount with virtual / zpower reading higher. I might do a 3 way test with my PM, TR and Zwift and graph the results.

Something to note, I could regularly hit > 1000W for 10+ secs on Zwift on zpower, in the 4 months I've had my PM I've only managed a max 1s power of 850W. virtual / zpower can flatter!

thomasmorris

Well, obviously one is wrong and quite probably both. But as both are based upon a power curve for the same model of trainer as yours it's impossible to say which.

Your ten mile tt time won't tell you much either. The best way to estimate power is from a long climb with a gradient over 5%.

I had a peek at you longest climb on strava, and saw an effort up the telegraph (better than Galibier as the altitude is less and it's around an hour long).
That VAM gives you a W/kg of about 4 (cycling power labs : with error as I don't know wind direction and am estimating Crr, CdA and bike weight). Looking at your heart rate it looks like a pretty hard effort (only 1 bpm lower than your trainer road FTP test).

Your trainer road says 3.6 w/kg for the FTP test, and zwift says over 5 w/kg for 13 minutes (which scaled down to ftp would be ~4.5 w/kg.

...assuming your telegraph performance was close to maximum I'd say your FTP is in the middle (~320 w). Incidentally, Strava estimated 318 watts for that telegraph climb as well.

However, as long as it's consistent it doesn't really matter for training, it just makes comparing backwards and to other people a bit woolly.

P.S. epic ride in the Alps, chapeau.

cgfw201

thomasmorris wrote:

Well, obviously one is wrong and quite probably both. But as both are based upon a power curve for the same model of trainer as yours it's impossible to say which.

Your ten mile tt time won't tell you much either. The best way to estimate power is from a long climb with a gradient over 5%.

I had a peek at you longest climb on strava, and saw an effort up the telegraph (better than Galibier as the altitude is less and it's around an hour long).
That VAM gives you a W/kg of about 4 (cycling power labs : with error as I don't know wind direction and am estimating Crr, CdA and bike weight). Looking at your heart rate it looks like a pretty hard effort (only 1 bpm lower than your trainer road FTP test).

Your trainer road says 3.6 w/kg for the FTP test, and zwift says over 5 w/kg for 13 minutes (which scaled down to ftp would be ~4.5 w/kg.

...assuming your telegraph performance was close to maximum I'd say your FTP is in the middle (~320 w). Incidentally, Strava estimated 318 watts for that telegraph climb as well.

However, as long as it's consistent it doesn't really matter for training, it just makes comparing backwards and to other people a bit woolly.

P.S. epic ride in the Alps, chapeau.

haha. excellent Strava stalking.

The Telegraph effort was conservative. First alpine climb and had no idea how I'd find it so consciously span up without going too deep.

The only climb I really had a crack at that week was Col de la Forclaz near Annecy,

https://www.strava.com/activities/69075 ... 6929053348

Not sure if that tells you anything different.

okgo

What do you weigh?

Someone I know who is about 80kg did the same time as you up that climb with 310w.

thomasmorris

Well, on Col de la Forclaz your VAM is higher, but the average gradient is steeper, so it gives a w/kg of 3.4, or 275 watts. The climb has downhill sections though which ruins my very basic calculations.

The strava estimate for this climb (311) is probably more on the money than my calculation as it takes into consideration the downhill sections. However, really it's not a great climb for this, so I wouldn't take anything from those numbers.

I think the 320 watt up telegraphe is pretty spot on (+- 10 watt). You could maybe add on 5 watts for pacing if you found it okay. The only way to get a better estimate than that is to to go out and do a harder effort, or borrow a power meter (I still see some questionable numbers from actual PM though).

Hopefully that's given you a rough idea.

stueys

Both zPower on Zwift and virtual power on TR are calculations, you only need one variable to be different to end up with very different results. I've used TR for years and know that the guys actually bought a load of turbo trainers which they painstakingly built and tested resistance curves on. Often they found issues with the manufacturer provided curves and created an amended version. But ultimately there is a lot of variability in friction provided by a turbo, especially if it's not a direct drive design.

Only real way to know which is more accurate is to test with a power meter. A power meter will also not temp drift.

markhewitt1978

Your are also likely to get quite different readings after your tyre is up to temperature

cgfw201

thomasmorris wrote:

Well, on Col de la Forclaz your VAM is higher, but the average gradient is steeper, so it gives a w/kg of 3.4, or 275 watts. The climb has downhill sections though which ruins my very basic calculations.

The strava estimate for this climb (311) is probably more on the money than my calculation as it takes into consideration the downhill sections. However, really it's not a great climb for this, so I wouldn't take anything from those numbers.

I think the 320 watt up telegraphe is pretty spot on (+- 10 watt). You could maybe add on 5 watts for pacing if you found it okay. The only way to get a better estimate than that is to to go out and do a harder effort, or borrow a power meter (I still see some questionable numbers from actual PM though).

Hopefully that's given you a rough idea.

nice one, cheers.

davidof

As others have said, there are a whole range of things that can create big differences. First of all Zwift and TR may be using different power curves for the same trainer. The Elite trainer has a cubic type curve, so how well do the equations that calculate power fit the curve? Has someone made a mistake?

However looking at the graph below I would suggest it is a fluid issue. A fluid trainer's power curve varies as the fluid warms (it gets easier to spin the paddles through the oil). This test against a real power meter suggests the Zwift curve is 25% optimistic once the fluid is warm (for the elite fluid red roller). The power curve is good round the 10 minute mark and at low power outputs.



TR have been doing this for longer so I would tend to trust their figures. They get a lot of user feedback about power. I suspect their power curve is based on a warmed up trainer.

Even given all that lots of other factors can have an effect on virtual power: tire, tire pressure, ride weight. You can do things like roll-down tests to get more consistency ride to ride but it is rarely worthwhile, IMHO.

amaferanga

I used to have an Elite elastogel fluid trainer and I can confidently say that the resistance is so tempetature dependent that trying to estimate power with it would be a total waste of time. If mine was normal then I'd suggest giving up on estimated power with that turbo. Try something like a Cycleops Jet Fluid or Fluid 2 if you want some degree of repeatability.

Constant(ish) power with Elite fluid trainer (note increasing speed):

https://goo.gl/photos/oZJvdf78rPiuPNfB9

Constant(ish) power with Cycleops Fluid 2:

https://goo.gl/photos/p84qUQBiDx9RLpVp8

Power for both was from a Power tap.

bianchimoon

amaferanga wrote:

I used to have an Elite elastogel fluid trainer and I can confidently say that the resistance is so tempetature dependent that trying to estimate power with it would be a total waste of time. If mine was normal then I'd suggest giving up on estimated power with that turbo. Try something like a Cycleops Jet Fluid or Fluid 2 if you want some degree of repeatability.

Constant(ish) power with Elite fluid trainer (note increasing speed):

https://goo.gl/photos/oZJvdf78rPiuPNfB9

Constant(ish) power with Cycleops Fluid 2:

https://goo.gl/photos/p84qUQBiDx9RLpVp8

Power for both was from a Power tap.

Very interesting to see difference between the 2 non smart trainers, have you ever used zwift with Zpower to quantify real life v virtual power difference?
As a Zpower user would love to know how close zwift actually comes to the real thing?
PS Sorry, realise i have hijacked thread if you do reply, could you do on 'Zwift on a non smart trainer thread'

okgo

With powermeters being the price they are now, I can't see why anyone is interested in playing bullshit bingo with these trainers and estimates?

markhewitt1978

okgo wrote:

With powermeters being the price they are now, I can't see why anyone is interested in playing bullshit bingo with these trainers and estimates?

Power meters are still several hundred. More to the point you can get a smart trainer - Tacx Flow - for £200 so you may all well get that

cgfw201

okgo wrote:

With powermeters being the price they are now, I can't see why anyone is interested in playing bullshit bingo with these trainers and estimates?

forgive me for not having £500 hidden down the back of the sofa

bianchimoon

okgo wrote:

With powermeters being the price they are now, I can't see why anyone is interested in playing bullshit bingo with these trainers and estimates?

Purely out of interest, i'm not racing or training for a race, I don't need a PM per se, but I do use zwift and the power output real or virtual is part of using the programme. Would be nice to know how close to reality it is without spending £500!

okgo

So rent one, they're peanuts to rent for a few months.

bianchimoon

okgo wrote:

So rent one, they're peanuts to rent for a few months.

I'll just leave it that zwift have calculated it as well as possible, thanks anyway

okgo

Well they haven't, because its all totally fictional and results in people coming on here asking whether they really can beat Cancellara in a time trial because Z power said X...

bianchimoon

okgo wrote:

Well they haven't, because its all totally fictional and results in people coming on here asking whether they really can beat Cancellara in a time trial because Z power said X...

Ouch, you're a bit angry today, i guess what you're saying is my zpower 1.5w/k is fictional and i'm really a 4.5w/k?
Better get back on zwift and knock my weight down to 60 kilos.
BTW educating people and spreading the word about correct use of Zpower using supported trainer, correct height/weight/ wheel size and tyre pressures would have a far greater effect on a level playing field for all... and zwift is just a game when all said and done

okgo

If its such a game why do people come on here asking how it compares to real life 1000 times a day?

As I said, if you want to know how good or bad you are, rent/buy a powermeter, if you want to play virtual top trumps, Zwift away...

stueys

I used trainer road and a magnetic trainer for about a year before I got a power meter. Ultimately I think virtual power is a good indicator to train with as it can be fairly consistent (if you are absolutely religious about set up (tyre pressure, trainer tension, etc, etc). But it's probably only consistent within itself, I suspect trying to compare against power meters and other trainers (using the same maths with multiple potential variables) is likely to be all over the shop.

For interest I compared TR virtual power against my power meter on a couple of rides. It wasn't very scientific but I found the readings were fairly accurate within a relatively narrow band but drifted further away as the power went up. Rough and ready assessment, I basically rode looking at my gamin compared to the TR display, take with a large pinch of salt.

amaferanga

bianchimoon wrote:

Very interesting to see difference between the 2 non smart trainers, have you ever used zwift with Zpower to quantify real life v virtual power difference?
As a Zpower user would love to know how close zwift actually comes to the real thing?
PS Sorry, realise i have hijacked thread if you do reply, could you do on 'Zwift on a non smart trainer thread'

Sorry, I don't use Zwift nor any other power estimating software (no PC capable of running it).