Kurt Kinetic Power Curve – VirtualPower

This is the first post in a series where we’ll start to look closer at the power curves of certain trainers.

Today, we’re going to check out the Kurt Kinetic fluid trainers.

There’s four models that I know of.
Kurt Kinetic Road Machine (blue – 3lbs fly wheel – not manufactured anymore)
Kurt Kinetic Road Machine (green – 6lbs fly wheel)
Kurt Kinetic Rock n Roll (green – 6lbs fly wheel)
Kurt Kinetic Road Machine/RocknRoll + Pro Fly wheel (green – 12lbs fly wheel)

As far as I know, they all have the same resistance unit.

It’s a fluid resistance unit with a sealed chamber.  They use a special fluid that doesn’t change in viscosity as it heats up.  This is what makes it kick ass for VirtualPower.

The sealed chamber has a disc in it. When that disc moves, it has to overcome the resistance of the fluid. That’s what creates the resistance while you workout.  The nifty thing they do is that the fluid chamber is completely sealed. They use magnets to connect the drum that you’re moving and the disc in the fluid. That’s how they guarantee that it will never leak.

So what does this all mean for VirtualPower?  Because their fluid doesn’t change in viscosity when it heats up you get a very consistent power curve during your workout.

Kurt Kinetic Power Curve vs PowerTap (no offset)

Here’s a workout I did with my PowerTap, which is a power meter that claims +- 1.5% accuracy.  The blue line is the PowerTap power and the red line is our VirtualPower which is calculated from speed.

For this setup I used a continental trainer tire at 120psi.  I cranked down my rear wheel 1.5 turns.

You can see that VirtualPower tracks really well with actual power.  Although VirtualPower is higher the entire ride.

This due to rolling resistance between the tire and the Kurt Kinetic drum.  We call this difference the y-intercept offset.

Kurt Kinetic Power Curve vs PowerTap (Adjusted Offset)

I threw my data into excel and changed the y-intercept offset by 30 watts.  When I did that, BOOM, VirtualPower and Power lock in.  There’s a little bit of difference in the first 5 minutes which I think is the rear tire heating up.  But after that it’s really close.  So close that the +-1.5% accuracy of the PowerTap makes these results even more compelling.

What does this tell us?

This tells us the Kurt Kinetic Road Machine works really well with VirtualPower.  I did the above test with the 3lbs fly wheel (blue model).  I talked to Kurt and they tell us that the power curve doesn’t change between the 3lb, 6lb and 12lb models.

If you’re holding a constant output, this makes sense.  You’re going to see some differences accelerating and decelerating. Because the 12lbs fly wheel has more mass it will take more work to get it spinning and it should stay spinning longer.  As long as you’re doing intervals of any considerable length you should be good to go with any configuration.

But….we’re still going to test the other fly wheels once we get our hands on them and publish them to the blog.  We like to double check our work ;).

So how tight should I get my rear tire?

I know what you’re thinking.  “Sweet, so you guys should be able to tell me how tight my rear wheel should be and I’ll be able to dial it in and have numbers that I can compare to power meters!”.

I can’t recommend that at the moment…here’s why.

It’s going to really depend on your tire, the tire wear and the PSI of your tires.  Tires have different rolling resistance and even with the same tire it changes over time.  I think as long as you use the same tire, pump it up to the same pressure and tighten the rear wheel down the same you’ll be able to use VirtualPower for training and get consistent power numbers.  But I still wouldn’t use it to compare to other people, even if everyone tightened down their rear wheel two turns.

What about calibration?

We’re experimenting with roll down tests.  With that, we’d have you spin up to 25 mph and coast to a stop. We’d record how long it takes to go from 20 mph to 10 mph.

The problem with that is ANT only transmits your data every wheel revolution.  Although this doesn’t seem like much you can get a pretty big variation if you time it just when it passes 20 and 10 mph.

We haven’t given up hope yet though.  You can collect all the points between 22 and 8 mph and predict where it passed 20 and 10 mph.  With that we hope we can really dial in calibration and WE can add that y-intercept offset to the VirtualPower formula on the fly for you.

That’s pretty exciting because we could account for heat drift during a workout that happens in other brands of trainers.

Next Steps

We’re in the processing of purchasing a motor to hook up to a stationary bike.  With this we plan on doing repeatable tests to develop better power curves and help us with our calibration numbers.

The one problem we’re running into is how to hook the motor up to the cranks/bottom bracket.  Any machinists out there?  If you have any ideas let us know in the comments.


Hard Work Pays Off

Put in the work now, and prepare yourself with TrainerRoad.

or learn more about TrainerRoad

Share this Post

Nate Pearson

Nate Pearson is the co-founder of TrainerRoad. He is an avid triathlete and cyclist, husband and father of two. His training is fueled by great coffee, BBQ and pie.

50 comments on Kurt Kinetic Power Curve – VirtualPower

  1. Ugo says:

    Just bought a KK Road Machine 2 and I was wondering why the calibration process was necessary in order to have absolute power data, if an exact match between speed and power exists. Why a simple speed sensor wouldn’t be enough? After reading all the previous comments (thank you, btw), the pretty simple answer is: rolling resistance. It introduces a variable offset to the power vs speed curve, depending on several external factors (mainly tyre type and how tightly it is set up).
    I suppose that KK have worked out the relation between the spin down time SdT (between 2 fixed speeds, 20mp/h and 10 mp/h) and the offset OFF: OFF(SdT), which I guess is linear. I wonder if KK are willing to provide that relation as well.

  2. Chastity says:

    CDs are generally issued by banks and, in most cases, are insured by the
    Federal Deposit Insurance Corporation (FDIC) for up to $250,000*
    per depositor. They also helped parents of a 13-month-old girl when they were laid
    off from their jobs because of missing too much work to be with her.
    Then make the loan repayment amount the first priority in your budget.

  3. Maximo says:

    Wonderful blog! I found it while searching on Yahoo News.
    Do you have any tips on how to get listed in Yahoo News?
    I’ve been trying for a while but I never seem to get there!
    Many thanks

  4. Great blog here! Also your web site loads up very
    fast! What host are you using? Can I get your
    affiliate link to your host? I wish my web site loaded up as quickly as yours lol

  5. I just like the valuable info you supply on your articles.

    I will bookmark your weblog and take a look at once more here frequently.
    I am slightly certain I will be informed many new stuff proper here!
    Best of luck for the next!

    Visit my site :: kabel (travel2siam.ru)

  6. google says:

    Yes! Finally something about google translate sings.

  7. Mick Jordan says:

    What’s the current state of play on getting the offset automated out? I ask because my Kurt Kinetic InRide has just given up the ghost and so I’m on VirtualPower until I get a replacement.

  8. Aw, this was a very nice post. Finding the time and actual effort to generate a really good article… but what can I say… I put things
    off a lot and don’t manage to get nearly anything done.

  9. […] In our unscientific tests with our PowerTaps and Kurt Kinetic Road Machines we were within 3 avg watts for our intervals…so pretty darn accurate. You can get a run down on the Kurt Kinetic Virtual Power Curve here. […]

  10. Crankin Stein says:

    Why don’t you just fit the resistance coefficients c1 and c3 so that the squared error between the recorded and calculated coast down curve is smallest.

    I already tried that and it worked pretty well. I got values of c1 = 7.12 N and

    c3 = 0.243 kg/m

    Translation equivalent is 15.2 kg without fly wheel pro and 49.8 kg with fly wheel pro.

  11. Peter says:

    this may be a stupid question about turns. why does it matter if its 1.5 or 3 or 5 turns as long as the tire does not slip?

  12. gray says:

    Doesn’t sound like the ant+ limitation of one data point a revolution should be too big a problem. Reckon you’d get about 30 or so data points out of a spin down down from 25 to 10mph which should be plenty to do some accurate very maths on. I’d certainly find this a very useful addition to the trainerroad arsenal. Maybe you could see if a keen mathematician amongst the trainerroad community could help you if you’re time limited!

  13. kingofbanff says:

    why do they suggest 1.5 turns above and all the threads talk about 5 turns?

  14. […] but if you don’t have one don’t worry: TrainerRoad’s VirtualPower is an incredibly accurate substitute. To use VirtualPower all you need is a compatible trainer, an ANT+ USB stick and an ANT+ […]

  15. jr says:

    Hi guys. On the rolldown test: the Ant sensor doesn’t sense rotation. It senses a magnet passing. If you want a better accuracy, put more magnets on the wheel, at regular intervals. Divide ant speed by number of magnets to get the exact speed….

  16. Joe Herrick says:

    Yes, exactly.  The unknown rolling resistance is the whole key.  I may be completely missing something, but how do you publish a power curve and then say set your resistance 2-5 turns?  The more I think about this, it seems like you need to use a number of turns very close to 5 to get actual power regardless of the type of tire you have.  I wonder if they just don’t want people to operate the trainer with that high of a resistance in fear of blowing a tire or damaging the trainer?  Or, why not bump the whole curve up by 20% or whatever the number is and then say to use 3 turns?

  17. Paul says:

    I’m not sure that I understand one part of this. If virtual power is too high then that means that the wheel is going around faster than it should be for that amount of power, according to the calculation. So by “this is because of rolling resistance” do you mean that rolling resistance is lower than assumed in the formula?

  18. Joe Herrick says:


    I believe that KK does not include the y offset because it is a function of an individuals tire, environmental conditions, and tension setting.  It is interesting that they publish this formula and then tell you to set your resistance between 2 to 5 turns.  2-5 Turns on the KK give the type of variation in virtual power data  as compared to measured power data that we are seeing.

    I think they key point that KK would say is that your virtual power reading with any device that calcualtes power based on speed is exactly proportional to your actual power reading as measured on powertap, etc. and therefore can practically be used for training purposes.  And if I interpret what Trainer Road’s position is, this is exactly what they are saying as well.

    I agree with your point that setting the tension correctly is the key to this whole issue.  And I agree that we all want the easiest way to get us the most accurate absolute power reading.

    It is interesting we both came up with five turns.  I did a very informal test of just riding along at about 150 Watts according to my powertap computer on my handlebars and glanced back and forth between my powertap Watts and my PC where my Trainer Road Virtual power was displayed.  I started out with 2 turns on my KK Road Machine and kept getting off my bike and turning my knob tighter one turn at a time until at about 5 turns my Watts finally matched up.  Then I varied my power between 100 and 200 Watts and sure enogh Trainer Road Virtual Power and PowerTap were giving almost identical readings throughout that whole wattage range within say a few Watts in either direction.

    It’s interesting because Computrainer uses a roll down test, but your tension needs to be in a certian range so essentially they do both a rough tension setting and then a y offset correction as well.

    We’ll see how Trainer Road responds, but I believe they actually generate a calibration curve for each trainer at a few points and then interpolate to get their number vs using an empirical formula.  Maybe, they can comment further on this?  But, again, I don’t think the key element are in the fine details of the formula, but instead determining this offset number that needs to be applied to the formula at the end.  Just my opinion. 

    1. Harry Roberts says:

      They did respond.  I just missed it.  You might have too but here’s what they said about the formula.

      Hey Harry, we’re using P = (5.244820) * S + (0.019168) * S3.  We talked to Kurt directly and they told us that’s the correct one.Yes, I think my force was too low.  As someone else said, 5 turns matched up with them.  It will be tire specific though. 

  19. Harry Roberts says:

    Joey one of the points I wanted to make was that KK’s own formula for calculating power for their units does NOT require a Y-Intercept.

    I agree that some type of roll down test might help people set the tension on their KK correctly.  I personally think using the KK formula and having the tension set correctly would make me happier with the way virtual power is calculated.  

    The KK should produce accurate power numbers provided the tension is set correctly.  The Million dollar question is how can we do that with the least amount of effort.

    I thought it was interesting that you too had to turn the knob 5 turns to achieve the correct tension and you used a completely different tire.

    The computer is simply KK’s device that uses speed and I assume their formula to calculate power. More specifically I think the corrected formula that I provided in my previous post is actually used to calculate power.  

    I guess when it’s all said and done it really doesn’t matter how it works (setting tension correctly or compensating for tension via y-intercept) as long as it works.

    I’d love to know what formula you guys are currently using.  Since KK provides a formula and the internet provided what most people believe to be the corrected version of the formula I’m wondering how close they are to yours provided you guys developed it yourself.

  20. Nate Pearson says:

    Thanks.  We also have a good machinist in Reno.  Kurt is going to tell us how to make that bracket.  We just need the time and money to put it all together :)

  21. Nate Pearson says:

    It’s on the back burner at the moment.  We’re waiting to get our motor bought and a bracket to connect to a bike.  I see us getting this done this summer.

  22. Nate Pearson says:

    Yes, we will for sure.  We just want to hook that motor up.  So busy…I say we do this in the summer.

  23. Joey Herrick says:

    I think you guys are missing the simple connection between the “y-intercept” and the number of turns on the knob, your tire, and your power meter.

    Of course if you have some independent way of measuring power, you simply adjust the number of turns until your virtual power matches your power meter numbers.  I have done this with a wired powertap system and a KK Road Machine as well and came up with about 5 turns on an old Vittoria Zaffiro tire (not trainer specific).  And if you have a way of directly measring power you don’t really have a need to use the virtual power feature.

    The whole point is that many people don’t have an independent measure of power so how many turns do they do for their set-up?

    Also, your tire tread is constantly wearing down and your tire pressure and temperature is not exactly the same every time your ride.  Also, your bike is not positioned in the trainer exactly the same way every time you ride.  Computrainer has you perform a “roll down” test each session to account for all these variables.  This test appears to be the “gold standard” for calibrating power for each session when using a trainer and calculating power as a function of speed.  This is what determining the “y-intercept” is all about.  Harry, you essentially determined your “y-intercept” for your setup that day by using your KK computer although I am not familiar with these and do not know how accurate they are.

    I don’t think it’s an issue of the formula per say as the Trainer Road guys have shown the curves essentially overlay perfectly if you scale the virtual power curve correctly (multiply everything by some unknown constant which they are referring to as the y-intercept).

    In addition to the amazing software that Trainer Road has developed, it is quite amazing they are very close to providing very accurate power based training without any additional equipment but your bike and a relatively inexpensive resistance trainer.

  24. Harry Roberts says:

    Just want to add my KK RoadMachine 1st use observation.

    I rode it simply to see if the wattage was accurate.  By accurate I wanted to see if it matched my SRM crank based power meter that I recently checked the calibration on.

    I am currently using a Tacx “Blue” trainer tire.  My conti tire blew out a few days ago.  This tire is specifically designed to be used on a trainer and I had it inflated to 120 psi.

    I also have the KK computer so that I can see the wattage KK would report.  That simply means that I don’t have to look up my speed in a chart to see what wattage KK would say I’m generating.

    I followed the KK instructions to tighten the knob 3-5 turns after the roller contacts the tire.  On my first attempts I went with 3 turns.

    I did a very simple test.  I rode short intervals (1-2 minutes) at 150, 160, 170, 180, 190 & 200 watts.

    From visually looking at the KK computer and my SRM I could tell that the KK reported wattage was high by about 10 watts.

    So I turned the know 2 more turns for a total of 5 and did one final test at 200 watts.  My SRM reported 206w average for that interval.  A crank based power meter like the SRM should read higher due to drive train losses than the KK trainer.

    So for me it appears that I need to tighten the knob 5 turns.

    No need to add a “Y-Intercept”.  Simply tighten the knob down correctly.

    What is correctly?  I have no idea unless you have another power meter to set the tension on the roller correctly.

    Possible sources of error.  My SRM may not be completely accurate.  I know I couldn’t hold the KK at a constant speed/wattage but do believe that I did a pretty good job.

    As a side note I tried using less tension that the 3 turns and the tire would slip on the roller at relatively low wattage.  

  25. Harry Roberts says:

    What formula are you using to calculate power for the KK?

    KK has this formula all over their website:  P = (5.244820) * S + (0.01968) * S3

    The problem is many are saying that’s not the correct formula and the actual formula should be:   P = (5.244820) * S + (0.019168) * S3

    The difference being the addition of a “1” between the 9 and 6 on the multiplier for speed cubed.

    There is support for the latter formula being correct.  KK also provides a chart of speed vs power and their formula does NOT match the charted data.  The latter formula matches the charted data.

    In either of the formulas above I do not see them using a “Y” intercept.  There is not a value added to the calculation to account for Rolling Resistance, press on force or whatever we might decide to call it.

    Is it possible the force you used on the roller against your tire was simply too low and accounted for the difference in wattage instead of needing to add a “Y” intercept?

    KK Road Machine instructions say to tighten it 2 to 5 turns after roller touches tire.

    1. Nate Pearson says:

      Hey Harry, we’re using
      P = (5.244820) * S + (0.019168) * S3.  We talked to Kurt directly and they told us that’s the correct one.

      Yes, I think my force was too low.  As someone else said, 5 turns matched up with them.  It will be tire specific though.

  26. Dxstout says:

    Great info guys! As to connecting a motor to the cranks, take a look at Kurt’s site. There they have a video and talk about their testing procedure, which involves an electric motor attached to the bike. Might be a simple copy job.

  27. dgilson says:

    Are you guys still in search of mechanical input for the motor/crank setup? 

    1. Nate Pearson says:

      Yes, Kurt Kinetic is actually going to help us with it.  We’ve just been super busy and haven’t done it yet.  I see myself getting on top of this during the summer.

  28. Guest says:

    Tried the same test on my trainer and found the coast down time for 20 mph, 2 turns, 110 psi was 13.6 sec. Very close to what VW005 found. For 1.5 turns time was 15.2 sec, which is about a 10% change.

    That would suggest that the agreement between powertap and the road machine should certainly be better if 2 turns are used.

    Seems like the problem with doing roll down tests for any kind of calibration goes beyond tire pressure/type. It must also be a function of the moment of inertia of the entire wheel. A lighter wheel set (although mass isn’t the only factor) will store less energy at a given speed and will roll down faster since there is less energy to dissipate by the trainer. The question is how significant this might be???.

    1. RenoJoe says:

      You are correct that the weight of the wheel does effect the roll down time which is why the roll down test is ideal because it does measure this effect.  This is also the reason why just calibrating to a certian number of turns can’t be applied universally to all wheels.  For example, it is obvious that calibrating to two turns on a 10lb wheel would be vastly different than calibrating two turns on a 1 lb wheel.  The turn calibration method does not take into account different wheel weight while the roll down method does which you pointed out.

      Theoretically you may be able to calibrate to a turn number, but it would only work for a specific wheel, with a specific tire, with a specific amount of tread wear at a specific temperature and pressure.

      I just ordered a powertap system and I am going to try this on my bike next week.  I’m also going to cross calibrate the powertap with an inexpensive digital luggage hanging scale ($10 on Amazon).  I am thinking I may be able to establish a force (scale reading) that it takes to slowly pull my wheel around by connecting the scale to a spoke that will give the same reading on trainerroad as powertap.  Then, maybe I can just use my digital scale to determine how tight to turn my trainer (kinetic) wheel against my tire each time I ride?  This could theoretically be applied to different different wheels on any kinetic trainer.  Or, maybe I just wasted 10 bucks.

  29. Anonymous says:

    I’ve been using 2 turns and speeding up to 20mph with a coast down time of about 13.5 seconds with the tire at 110 psi.  What about a study from your powertap about how much each half turn increases your wattage by?  It would be good to know if the variance is significant or in the noise.

    1. RenoJoe says:

      How are you measuring your roll down time?  I am waiting for my powertap to arrive and I will look at wattage change vs turn amount and let you know what I find.

      1. Anonymous says:

        That would be great! Im doing it in a very unscientific way. Im just turning on the Garmin and iPhone stopwatch. Once I get in a steady state speed of 20mph on the Garmin, I stop pedaling and hit the stopwatch start. Then I look down at the wheel and stop the watch the moment the wheel stops. I figure my roll time would be accurate within .5 seconds.

        1. Nate Pearson says:

          We’re going to do roll down time from 20 to 10 mph using ANT.  We’re trying to get super accurate though and we need to do some math to interpolate the points.

  30. arc99 says:

    Any chance you can do a similar analysis for the CycleOps Fluid2?

  31. Guest says:

    I just purchased a road machine and am interested in the power/speed relationship for training purposes. I noticed that the current DVD that comes with the machine says to tighten the rear wheel down 2 full turns after the roller just makes contact with the tire. I would expect that going from 1.5 to 2 turns might push things into closer agreement.

  32. Ieris says:

    How things are going with calibration? Waiting for this featrue…

  33. Dan Van Drunen says:

    Nate, my brother-in-law is an excellent machinist if you are still looking for some help getting your motor rigged up. Drop me a line and I can connect you and him. FYI, he is not a cyclist though so you would need to explain bike components to him. 

  34. Anonymous says:

    How tight the back wheel needs to be to the roller is the reason why I prefer my Elite Hydromag because that issue never crops up….. the bike rests on the roller using rider weight to determine the resistance, the same as on the road…… My assumption is that this is a better setup?

    1. Nate Pearson says:

      That should keep rolling resistance consistent…as long as you don’t change weight :).

      1. Anonymous says:

        Providing I don’t lose whole limbs on a ride then weight fluctuation between rides shouldn’t be noticeable in terms of rolling resistance :D

  35. lindy says:

    Have your considered a tandem crank for the stoker and just use a standard bike chain from the motor to the left side of the crank?  Or go to the inner or outer ring of a triple crank.
      An electric scoot motor already comes with the chain cog on the motor  (e.g. http://www.electricscooterparts.com/motors60volt.html).  

    1. Nate Pearson says:

      We thought about running our own chain on the big or small ring and having the standard chain run the rear cog.  I like that idea of using an electric scooter motor though!

  36. Akennedy Coas says:

    Rons Sines (look him up on Strava) is a machinist.  He might be able to make you a robojig for your bike. 

      1. Joey Herrick says:

        Nate,  Have you guys thought about placing multiple spoke magnets on the wheel to pick up more data points giving you higher velocity resolution during roll down calibration?  It would make the rear wheel unsightly, but magnets are cheap and readily available and many people have a dedicated trainer wheel and tire anyway.

        1. dgilson says:

          Agreed, magnets are cheap and easy to install for a quick calibration routine. 

Comments are closed.


Subscribe to the Blog

Join for the latest training, racing, and software updates from TrainerRoad.

Weekly Training Tips

Subscribe to the TrainerRoad Blog to receive articles on getting faster sent to your inbox!

No thanks, I'm fast enough.