Bicycle frame analysis on FEA
Heys guys, i'm doing a FEA on a road bike frame for a school work, which consists of a force applied on the driver's seat, being the frame fixed by the points represented in the image.
The objective is to test this for 4 materials, and to obtain for each the lightest frame by chosing several tubing, while not surpassing the maximum stress (elastic limit) and deflection allowable.
Does anyone have any idea how can i get an approximate minimum vertical stiffness (or maximum deflection for a given vertical force)? Is there a norm for this, or what values of vertical stiffness are considered normal?
There doesn't seem to be any tests regarding this no matter where i look. I've found on a website that there is a Seat Load/Vertical Rigidity Test, but a google search doesn't give me any good results. Any help is appreciated.
Thanks in advance !
The objective is to test this for 4 materials, and to obtain for each the lightest frame by chosing several tubing, while not surpassing the maximum stress (elastic limit) and deflection allowable.
Does anyone have any idea how can i get an approximate minimum vertical stiffness (or maximum deflection for a given vertical force)? Is there a norm for this, or what values of vertical stiffness are considered normal?
There doesn't seem to be any tests regarding this no matter where i look. I've found on a website that there is a Seat Load/Vertical Rigidity Test, but a google search doesn't give me any good results. Any help is appreciated.
Thanks in advance !
I found a bunch of stuff, I don't know if any of this is useful. I know many of the german magazines have testing methodologies for these kind of things.
https://www.giantbicycles.com/backoffi ... nWithGiant–RoadFrameTestData–FINAL.pdf
http://www.cervelo.com/en/engineering/t ... ality.html
http://blog.velocitebikes.com/2013/04/ ... 2hPMq.dpbs
http://www.sheldonbrown.com/rinard/EFBe ... e_test.htm
http://cozybeehive.blogspot.ca/2010/01/ ... mbers.html
http://www.tourmagazin.de/services/qtr ... age85.html
https://www.giantbicycles.com/backoffi ... nWithGiant–RoadFrameTestData–FINAL.pdf
http://www.cervelo.com/en/engineering/t ... ality.html
http://blog.velocitebikes.com/2013/04/ ... 2hPMq.dpbs
http://www.sheldonbrown.com/rinard/EFBe ... e_test.htm
http://cozybeehive.blogspot.ca/2010/01/ ... mbers.html
http://www.tourmagazin.de/services/qtr ... age85.html
I did something similar some 15 years ago. But it was more to verify my load data/points to some previously published papers on another study so that I may apply the verified methodology for my undegrad thesis paper.
For one thing, because of how simple and optimised a bike frame is, if your details of modelling such as angles, joints and wall thickness are not accurate, you'll not get any results close to what's 'real world', don't mind about performance criteria as you're talking about. In short, "rubbish in, rubbish out"...
i) Get your model as accurate as possible. Including wall thicknesses. (Note different materials have different thicknesses, as that's what out in the 'real world'. If its all the same, then the value of the study diminishes dramatically. You can either alter your models to reflect identical geometry or identical mass to simplify things.
ii) For your reference, you can't look at deflection/stiffness per tube section, but at the overall von mises stress at key points such as below the downtube junction with headtube, the BB intersection and see how it references back to the material's UTS , Yield Strength, fatigue cycle tables.
iii) as for overall deflection, you'll can refer to the Tour Magazine figures, they do publish their testing methodology and see how you can extrapolate your data from there.
And finally ... good luck! It can be hair pulling at best of times.
Edit: Forgot to add 
your load points are not correct. Because load is applied both at the saddle and at the handlebar.
i) Due to saddle offset, load at the seattube joint you pointed will consist of a moment and a load.
ii) The will be a vertical load + a moment at the headtube juncture too.
iii) due to the angle and offset rake of the fork, your headtube junction cannot be "fixed"if your rear axle is fixed. This is because of a shear motion induced by the angle.
iv) Also, there should be an effective moment load at the BB axis too. This is due to the twisting forces induced on the frame from your pedaling force. This is assuming you are inducing a fixed sidetoside movement at the 2 wheel axles for your frame loading parameters.
Bear in mind, all these are from memory of work done all those years ago. Not all the details may be correct but generally, the modeling of loads/loading is a whole lot more complicated than what you're inferring here.
For one thing, because of how simple and optimised a bike frame is, if your details of modelling such as angles, joints and wall thickness are not accurate, you'll not get any results close to what's 'real world', don't mind about performance criteria as you're talking about. In short, "rubbish in, rubbish out"...
i) Get your model as accurate as possible. Including wall thicknesses. (Note different materials have different thicknesses, as that's what out in the 'real world'. If its all the same, then the value of the study diminishes dramatically. You can either alter your models to reflect identical geometry or identical mass to simplify things.
ii) For your reference, you can't look at deflection/stiffness per tube section, but at the overall von mises stress at key points such as below the downtube junction with headtube, the BB intersection and see how it references back to the material's UTS , Yield Strength, fatigue cycle tables.
iii) as for overall deflection, you'll can refer to the Tour Magazine figures, they do publish their testing methodology and see how you can extrapolate your data from there.
And finally ... good luck! It can be hair pulling at best of times.
Edit: Forgot to add 
your load points are not correct. Because load is applied both at the saddle and at the handlebar.
i) Due to saddle offset, load at the seattube joint you pointed will consist of a moment and a load.
ii) The will be a vertical load + a moment at the headtube juncture too.
iii) due to the angle and offset rake of the fork, your headtube junction cannot be "fixed"if your rear axle is fixed. This is because of a shear motion induced by the angle.
iv) Also, there should be an effective moment load at the BB axis too. This is due to the twisting forces induced on the frame from your pedaling force. This is assuming you are inducing a fixed sidetoside movement at the 2 wheel axles for your frame loading parameters.
Bear in mind, all these are from memory of work done all those years ago. Not all the details may be correct but generally, the modeling of loads/loading is a whole lot more complicated than what you're inferring here.
I believe the maximum stress would be when the rider is sprinting. So you need to add the rider's weight on one of the pedals and also his pull on the same side of the handlebars. The rider also pulls on the other pedal. Saddle bears no load in this example.
Also the chain compresses the right chainstay. Through the hub, a little bit of force is also transferred to the left chainstay.
Which FEA program are you using? I tried Patran with limited success, using the scenario above.
Sent from my iPhone using Tapatalk
Also the chain compresses the right chainstay. Through the hub, a little bit of force is also transferred to the left chainstay.
Which FEA program are you using? I tried Patran with limited success, using the scenario above.
Sent from my iPhone using Tapatalk
Last edited by efeballi on Mon Apr 14, 2014 10:33 am, edited 1 time in total.
SHUT UP LEGS
2015 Giant Propel Advanced
2013 KTM Strada mod. GOT IT BACK!
2011 Pinarello Dogma 60.1(loaner)
2011 Scott SUB 45(sold)
Politecnico di Milano Ingegneria Meccanica
2015 Giant Propel Advanced
2013 KTM Strada mod. GOT IT BACK!
2011 Pinarello Dogma 60.1(loaner)
2011 Scott SUB 45(sold)
Politecnico di Milano Ingegneria Meccanica

 Posts: 589
 Joined: Fri Feb 10, 2012 4:27 pm
Nice stuff there.
Fellow Engineer who works with FEA as well. I work with biomedical stuff though.
Maximum Deflection for a given load FOR a given material and a given tube cross section can be found by using the generalized euler's beam bending theory or beam buckling theories, depending on whether your tube is vertical or horizontal and your direction of force. But this will be for simple beams, mind. you will need to analyze each tube differently. there would be no 'normal' because it would depend on tube cross section, composite layup etc. there would be a 'range' of what bicycle makers would probably do, the ones with less deflection being your all out race bikes and those with more deflection being your enduracnce models. others have pointed that out already. Which is why we have FEA to do all this nasty legwork for us.
Will agree that your loading and boundary conditions seem to be incorrect. maxxevv has covered most of it. However as this is a school project I don't expect it to be a super complicated, so try to start with a simple simulation before you start adding forces. the general rule with FEA is that you start with a simple simulation with lots of general and possible inaccurate assumptions. when that works, refine the model and the associated assumptions.
Fellow Engineer who works with FEA as well. I work with biomedical stuff though.
Maximum Deflection for a given load FOR a given material and a given tube cross section can be found by using the generalized euler's beam bending theory or beam buckling theories, depending on whether your tube is vertical or horizontal and your direction of force. But this will be for simple beams, mind. you will need to analyze each tube differently. there would be no 'normal' because it would depend on tube cross section, composite layup etc. there would be a 'range' of what bicycle makers would probably do, the ones with less deflection being your all out race bikes and those with more deflection being your enduracnce models. others have pointed that out already. Which is why we have FEA to do all this nasty legwork for us.
Will agree that your loading and boundary conditions seem to be incorrect. maxxevv has covered most of it. However as this is a school project I don't expect it to be a super complicated, so try to start with a simple simulation before you start adding forces. the general rule with FEA is that you start with a simple simulation with lots of general and possible inaccurate assumptions. when that works, refine the model and the associated assumptions.
locktopus wrote:I found a bunch of stuff, I don't know if any of this is useful. I know many of the german magazines have testing methodologies for these kind of things.
https://www.giantbicycles.com/backoffi ... nWithGiant–RoadFrameTestData–FINAL.pdf
http://www.cervelo.com/en/engineering/t ... ality.html
http://blog.velocitebikes.com/2013/04/ ... 2hPMq.dpbs
http://www.sheldonbrown.com/rinard/EFBe ... e_test.htm
http://cozybeehive.blogspot.ca/2010/01/ ... mbers.html
http://www.tourmagazin.de/services/qtr ... age85.html
Hehe thanks m8 but i had already looked through almost all of those (first one seems to need password...) and they give some general information but nothing that i can use for here. Thanks though :p
maxxevv wrote:I did something similar some 15 years ago. But it was more to verify my load data/points to some previously published papers on another study so that I may apply the verified methodology for my undegrad thesis paper.
For one thing, because of how simple and optimised a bike frame is, if your details of modelling such as angles, joints and wall thickness are not accurate, you'll not get any results close to what's 'real world', don't mind about performance criteria as you're talking about. In short, "rubbish in, rubbish out"...
i) Get your model as accurate as possible. Including wall thicknesses. (Note different materials have different thicknesses, as that's what out in the 'real world'. If its all the same, then the value of the study diminishes dramatically. You can either alter your models to reflect identical geometry or identical mass to simplify things.
ii) For your reference, you can't look at deflection/stiffness per tube section, but at the overall von mises stress at key points such as below the downtube junction with headtube, the BB intersection and see how it references back to the material's UTS , Yield Strength, fatigue cycle tables.
iii) as for overall deflection, you'll can refer to the Tour Magazine figures, they do publish their testing methodology and see how you can extrapolate your data from there.
And finally ... good luck! It can be hair pulling at best of times.
Edit: Forgot to add 
your load points are not correct. Because load is applied both at the saddle and at the handlebar.
i) Due to saddle offset, load at the seattube joint you pointed will consist of a moment and a load.
ii) The will be a vertical load + a moment at the headtube juncture too.
iii) due to the angle and offset rake of the fork, your headtube junction cannot be "fixed"if your rear axle is fixed. This is because of a shear motion induced by the angle.
iv) Also, there should be an effective moment load at the BB axis too. This is due to the twisting forces induced on the frame from your pedaling force. This is assuming you are inducing a fixed sidetoside movement at the 2 wheel axles for your frame loading parameters.
Bear in mind, all these are from memory of work done all those years ago. Not all the details may be correct but generally, the modeling of loads/loading is a whole lot more complicated than what you're inferring here.
1) Maybe i should have said on the begginning, these boundary counditions are not for me to chose. The teacher defined these, and i have to use them...
2) Thanks for the tips on looking for the maximum stress and deflection. =) The maximum stress for the frame will be the yield point stress of the material (so said the teacher), about the maximum deflection he told us to figure out.
davidalone wrote:Nice stuff there.
Fellow Engineer who works with FEA as well. I work with biomedical stuff though.
Maximum Deflection for a given load FOR a given material and a given tube cross section can be found by using the generalized euler's beam bending theory or beam buckling theories, depending on whether your tube is vertical or horizontal and your direction of force. But this will be for simple beams, mind. you will need to analyze each tube differently. there would be no 'normal' because it would depend on tube cross section, composite layup etc. there would be a 'range' of what bicycle makers would probably do, the ones with less deflection being your all out race bikes and those with more deflection being your enduracnce models. others have pointed that out already. Which is why we have FEA to do all this nasty legwork for us.
Will agree that your loading and boundary conditions seem to be incorrect. maxxevv has covered most of it. However as this is a school project I don't expect it to be a super complicated, so try to start with a simple simulation before you start adding forces. the general rule with FEA is that you start with a simple simulation with lots of general and possible inaccurate assumptions. when that works, refine the model and the associated assumptions.
Im using ABAQUS. As for the rest, i don't think we're supposed to find out the maximum deflection that the tube can bend, but some value that is normal for this industry. It's just that from my research no one seems to do this test because it's basically close to useless, because it neither tests the most important stiffness (tornsional and lateral) and doesn't give a good idea about rider comfort (for which case it seems manufacturers seem to prefer to test with the fork and more of the bicycle elements included than just the frame).
Hence why it's being to hard to find some values for vertical stiffness/compliance/maximum vertical deflection for a test like this. And yes this is not supposed to be super duper realistic, just a start point, but still i need some base values.
From what i could see, some people talk about "vertical compliance" and i've seen values like for vertical stiffness of 200N/mm. But that would mean if i put a load of 100kg (heavy rider) it would bend 5cm which seems a lot if it was just the frame. But then again i think that's a value for the vertical stiffness of the whole bike not just the frame!
This is one of the results i got for an ordinary set of steel (reynolds 531) tubes available from a site that sells bike tubes
For F=120kg*4G=4800N (approx 440kg) which gives max vertical displacement of 0.09mm... isn't that too little??
I actually went ahead and tried 5ton and it just went down 2mm! (didn't reach yield stress (540MPa) anywhere on the frame... (max was 378MPa)
Thanks for your replies fellow engineers! I'll keep looking for some values to use
As said, rubbish in, rubbish out. That's FEA for you.
The boundary conditions and loads suggested by your teacher is really far off what the real world loads/boundaries would be, hence the results you're getting.
Also, looks like your model consists of solid bars and not hollow tubes ?
Modify them into hollow tubes if they aren't already so, the readings/ results will be radically different.
The boundary conditions and loads suggested by your teacher is really far off what the real world loads/boundaries would be, hence the results you're getting.
Also, looks like your model consists of solid bars and not hollow tubes ?
Modify them into hollow tubes if they aren't already so, the readings/ results will be radically different.
maxxevv wrote:As said, rubbish in, rubbish out. That's FEA for you.
The boundary conditions and loads suggested by your teacher is really far off what the real world loads/boundaries would be, hence the results you're getting.
Also, looks like your model consists of solid bars and not hollow tubes ?
Modify them into hollow tubes if they aren't already so, the readings/ results will be radically different.
Yeah, the fact that we're using useless and unrealistic boundary conditions would explain a lot...
The tubes are hollow. They're modeled as beam elements and then each beam was given a profile (tube section). ABAQUS doesn't show the view you see there in the stress image (by default it shows each tube as a line, like the displacement image).
 Mattias Hellöre
 in the industry
 Posts: 554
 Joined: Fri Sep 23, 2005 6:34 pm
 Location: Insjön, SWEDEN
 Contact:
No weight on BB? there should be anything so you get a more realistic load case but I don´t recommend FEA on a frame, take something simpler and more near the real life, like a crank arm, BB axle or something like it.
Experimental Prototype

 Posts: 589
 Joined: Fri Feb 10, 2012 4:27 pm
You confuse me. You say you're. Supposed to find benchmark values but for what location? Deflection of the different parts of the frame are obviously going to ne different depending on material and shape as well. So what exactly are you trying ti find?
You can try velonews magazines. In their bike reviews tthey do a frame stress test in a jig with measured deflection at certain places.
You can try velonews magazines. In their bike reviews tthey do a frame stress test in a jig with measured deflection at certain places.

 Posts: 589
 Joined: Fri Feb 10, 2012 4:27 pm
One more thing: what material are your tubes? If they are cf, youve got to define the layup direction properly. Huge difference. Comppsites modelling in fea is tough
The materials are Reynolds Steel 531, Aluminum True Temper T2, CF Standard UD, and Ti3Al2.5V.
I decided to test on ABAQUS, see the results and then decide which value to adopt for the maximum displacement of the seat point. Since this is an unrealistic test anyway, it's very hard to obtain a value a priori...
I put a load of 4800kN and used the same frame size for all the materials and got the displacement of the point of the seat.
Then i decided on a value not too big and not too small (0.32mm) that i could get to by using different tube sizes with all 4 materials. And got these results (the values for the tubes is the outer diammenter and not specified units are in mm)
The carbon fiber seemed to be the lightest by far, but of course this is not realistic (for one we're using UD CF, and that's not the case in bike frames). Aluminum frame is the biggest of them all, but is the one with smallest maximum stress.
Any criticism? I know some of them are probably not realistic, but i don't have much time, so i'll probably just run with these. Still i'll mention that on report.
I decided to test on ABAQUS, see the results and then decide which value to adopt for the maximum displacement of the seat point. Since this is an unrealistic test anyway, it's very hard to obtain a value a priori...
I put a load of 4800kN and used the same frame size for all the materials and got the displacement of the point of the seat.
Then i decided on a value not too big and not too small (0.32mm) that i could get to by using different tube sizes with all 4 materials. And got these results (the values for the tubes is the outer diammenter and not specified units are in mm)
The carbon fiber seemed to be the lightest by far, but of course this is not realistic (for one we're using UD CF, and that's not the case in bike frames). Aluminum frame is the biggest of them all, but is the one with smallest maximum stress.
Any criticism? I know some of them are probably not realistic, but i don't have much time, so i'll probably just run with these. Still i'll mention that on report.

 Similar Topics
 Replies
 Views
 Last post

 16 Replies
 2007 Views

Last post by sooni
Sun Jul 09, 2017 11:03 pm


Sram Red BB386 Crankset with BB386 Bottom Bracket in BB386 Frame: Small Chainring touches Frame
by jbattermann » Wed Jul 19, 2017 11:22 pm » in Road  5 Replies
 1095 Views

Last post by KCookie
Sun Jul 23, 2017 5:06 am


 23 Replies
 2974 Views

Last post by CarlosFerreiro
Thu Apr 06, 2017 11:21 am

 6 Replies
 1099 Views

Last post by addictR1
Fri Jul 14, 2017 5:32 pm

 3 Replies
 358 Views

Last post by cyclenutnz
Sun Oct 22, 2017 2:20 am