Quest frame flex tested

[billyk]

Back in April we discussed flex between the handlebars and the bottom bracket/pedals of the Quest (http://cruzbike.com/energy-sucking-handlebar-stick). Various people pointed to the handlebar bending, the Diatech headset, etc. John T noted that it is the accumulated flex of the whole system, and that the design of the Silvio/Vendetta is intended to reduce this.

Rich suggested that "As a test, a bar-end could be attached at the top of the riser (right below the handlebars) pointing forward (and angled down) and a pipe stuck on that sticking out forward to a point somewhere near the bottom bracket". A very good idea that I finally got a chance to try.

My setup (see photos) consists of
* The back end of a stem clamped on to the extension tube below the actual handlebar-holding stem. (Bar-ends don't fit on the extension tube)
* A custom mount to hold a 5/8-inch dowel (Rich's "pipe") pointed forward from the stem piece. This is from Brazilian Rosewood (masa randuba) which is an exceptionally hard wood that can be machined much like metal (with sharp tools), and can be drilled to make exact holes like metal. The mount bolts on to the stem and the dowel goes through it.
* A piece of clear polycarbonate is screwed into the front end of the dowel, with tickmarks filed into it. The ticks are 1/2-inch apart, roughly 1 degree at this distance.
* A vertical threaded rod sticks up from the BB, serving as a pointer against the tickmarks. (This rod is part of my fairing; see "new homemade fairing for Quest" http://cruzbike.com/new-homemade-fairing-quest)

This setup reveals any flex from the extension tube below the handlebar to the bottom bracket, separate from handlebar bend.

I've ridden this for about 25 miles, on all kinds of grades. The dowel is surprisingly rigid and it is quite easy to read the deflection of the vertical rod against the tickmarks.

Results expressed as observed *peak-to-peak* deflection in fractions of a tick. I don't claim accuracy of much better than half a tick:

- Riding no-hands: zero
- Maintaining steady speed of 15mph on a flat: up to one-half tick
- Normal acceleration from stop: up to 1 tick
- Hard acceleration from stop: up to 1.5 ticks
- Climbing a shallow grade (2-5%): 1-1.5 ticks
- Climbing a hard grade (about 8% per google maps): 2 ticks (more if in a too-high gear)
- Holding the front brake and pushing the pedal as hard as I could: 1.5 ticks <<== Each side separately (equivalent to 3 ticks of the others)

Thus there is a lot of flex between the extension tube and the BB, aside from anything to do with the handlebars. It makes little difference on the flat, but is a big deal when climbing hard.

To test the handlebar bend, I ran fine steel wire from the brake lever mounts to the end of the dowel, with each one fixed independently at both ends, just less than tight. These sensitively (if not quantitatively) show distance change between either handlebar and the dowel end. That is, if one side of the handlebar was bending back as I pulled, the wire would stiffen visibly. OK, definitely not quantitative but a yes/no test.
I repeated the previous tests. No change in the wire lengths was seen (or felt in tests with the brake held tight while pressing the pedal).

Thus, I conclude that most of the flex is in the boom/slider arrangement, including the pivot clamp at the top of the slider. Very little if any is in the handlebar. Agree?

Now, what to do about this?

I don't want a Silvio or Vendetta because I need the more upright seat of the Q for extensive mixing it up with car traffic, pedestrians, dogs, ..., plus being able to see potholes etc. (Can't afford one, anyway). I'm very happy with the Quest after more than 4000 miles; I just don't want to waste power when I'm climbing hills and need every Watt.

It seems like a longer slider would help, since I now know the length I need and do not adjust it. I seem to remember such being available but don't see it for sale on the web page.

The long slider could be further stiffened with another tight-fitting tube inserted inside it.

Opinions/suggestions welcome!

Billy K

 

[MrSteve]

Excellent!

Great job, brilliantly executed!
Thank you, thank you for sharing your result here.

(I know what my next modification will be.)

Did you note how thankful I am?

Thank you, Billy K.,

Sincerely,

-Steve

 

[richa]

Silvio 1.0 or 1.5


Very Impressive testing.

I'm actually surprised at the amount of deflection; I'd have guessed early on that most of the energy waste was going into twisting the bars. But it appears the flex in the rest of the structure is the biggest issue. Which explains the evolution of John's designs; a more direct connection between the BB and the handlebars (original Silvio), and then beefier connection (larger diameter tube) between the BB and handlebars (Vendetta). It's no surprise that the Vendetta is regarded as the best climber, since it wastes the least energy flexing the structure.

I'm curious how "hard" you pushed the pedal when holding the front brake and getting 1.5 ticks of deflection. It would seem to me that however much force you were using in that test, is the same amount of force (energy) being wasted in any other test that caused 1.5 ticks of deflection (climbing, accelerating hard). Which I'm envisioning is a lot. But this explains why I struggle up hills on the Quest that I can much more easily ride on my DF mountain bike. The design is clearly not optimal when it comes to energy use.

Unfortunately, I don't see any "fix" for an existing Quest. I see no simple "add-on" that would help; you need a more direct connection between the bars and BB and it needs to be solidly attached to both.

 

[Jeremy S]

Hi Billy, I could be wrong

Hi Billy, I could be wrong but I think that since you mounted your dowel high up below the handlebars, your results do not distinguish flex/twisting in the steerer extension tube from flex in the boom. I'm curious whether the results would improve at all by lowering the dowel on the steerer extension tube, since lowering the handlebars is a feasible mod for these bikes.

 

[billyk]

Right. Right.

Right. Rich asked about the 1.5-tick deflection from holding the brake and pushing the pedal hard. In fact that one alone is a single-side deflection, since I can only do it one side at a time. This, it is equivalent to 3 ticks of the others. This makes sense since I was pushing as hard as I can in a steady motion, maybe equivalent to riding uphill in too high a gear. Thanks for pointing this out. I have now edited the first post to reflect this.

Jeremy asks about moving the stem/dowel mount further down the extension tube. I'm away this week but will try that next weekend. It won't go down too far though, because the tube thickens above the headset. Maybe I can find a stem for a wide diameter? (This stuff depends on a great local shop that recycles bikes and has bins of used parts; a great resource). And I will have to sacrifice the logo placard. (First photo). But that would imply that the extension tube is twisting. I'd be surprised if that were the case.

Hoping to to hear John T's or Doug B's thoughts before making modifications.....

Billy K

 

[cllsjd]

I'm only a year behind, but I thought I'd ask a few question and see where I go. I'm not criticizing what was done. I've created a finite element model using beam element to model the front end of my Quest. That wasn't too hard. Then I needed to come up with some reasonable loads to apply to the model. I focused on the 5% slope and 1.5 ticks mentioned in the post by Billy K. If it were me, I'd have about 230 lbs (bike, rider, water, ....). In my dreams I might be using the 39 tooth chain ring and the 16 tooth sprocket on the cassette. I use clipless pedals putting the center of my foot about 5" from the frame center. Doing a little math I came up with loads to apply to my model. I ran the model and came up with about 3/8" of deflection at the bottom bracket. That seems to match up fairly well with the test. Billy, to better match my model to your test I'd like to know: how much weight was going up the hill; what gearing you might have used; finally, how far is it from the center of the frame to the center of your foot.

Assuming my model matches reality, I should be able to tweak the various component to lessen the deflection at the bottom bracket.

chuck

 

[billyk]

Hi Chuck - Well, I've been behind on this, too. I'd be glad if you can learn something more about this.

I hope you've seen Hamish Barker's model described in the thread "Riding Techniques/Incredibly stable recumbents", continued in the thread "Bicycle stability (2)". My post at the top of the second one links to two technical papers describing the model. That one is mostly about stability, but might be relevant here.

I weigh about 175lb, plus a few lbs of clothing/helmet/shoes. The plain bike weighs about 32lb. The fairing mounts (without fairing for this test) are light plastic, maybe 1lb, and I have a rack-bag with tools and miscellaneous junk, say another 5. Fenders, say 3lb? That would come to 220lb or so.

I have a 36-tooth round single chainring, and up a steep hill would be in lowest gear (34-tooth), and the internal hub in low range, a factor of 3/4, I think.

I'll measure the frame distance tomorrow. But what do you mean by "the center of the frame"?

Billy K

 

[cllsjd]

Sorry, my fault I didn't consider the internally geared hub. I really want to know the gear ratio. Shift the bike into the gear you would use to climb the hill. Then turn the crank arms / chain ring one complete turn and tell me how many revolutions the wheel made. You can put the bike on a stand or walk along side of it. Whatever is easy for you. My 39 tooth chain ring and 16 tooth "rear" sprocket would produce 2.438 turns of the wheel. Not that I expect you to give me an answer accurate to three places. I'd settle for an accuracy to a 1/4 of a turn.

For the center of the frame just measure from the center of one foot to the center of the other foot. Or a reasonable approximation of that distance. With clipless pedals I'm fixed at about 10". With platform pedals you could move your feet a little further apart. This would increase the moment applied at the bottom bracket and therefore the amount of deflection.

thanks for trying to help me out.

chuck

 

[billyk]

Hi Chuck -

The distance from the swingarm joint to the BB is 91cm, to the center of the pedals at full extension is 107cm (I have 155mm cranks). The swingarm joint is right under the bend of the seat, roughly my center of mass.

According to the Quest main page, it has:

"SRAM Dual Drive Internally Geared 3-speed hub, with 9-speed 11-28 derailleur system giving 27 gears. Crank length is 155mm. The Gear-inch range for the 26" model is 26.3 to 124.7"​

But this is with the standard 42-tooth elliptic e-ring. For my 36-tooth chainring, Sheldon Brown's gear-inch calculator says the range is 24.5 to 115.9.
(Those don't seem consistent, since changing the chainring should change both numbers by the same fraction).

Thus approximately, the gear I am in when doing hard hill-climbing is about 25 gear inches.

Since the SRAM internal hub range is 3/4 - 1.0 - 4/3, I think the rotation ratio you would use for my lowest gear is:

(36/28)*3/4=0.96 turns of the wheel for one rotation of the pedals.

Billy K

 

[cllsjd]

Billy, thank you for your help.

My finite element model seems to be producing reasonable result. In addition to the numbers you gave me, I did some of my own measurements. While my driveway isn't very long it is a 7.3% grade. Just holding the bike stationary produced about 3/4" of deflection. My model predicts .72 inches. I'll have to make a sketch with my nodes and coordinate system. Then I can publish the x,y, and z displacements at each of the nodes. Then I can increase the stiffness of each beam and see how the displacement changes.

chuck

 

[cllsjd]

I created a finite element model and ran some initial calculations. Let me make some disclaimers and realize I probably should have about twice the number I have.
This is an approximation. I measure my Quest to get the node location. Your bike would have different node locations; therefore, different results. My loads are an approximation
for me riding up 7.2 percent grade in a really big gear (53T chainring 16T cassette sprocket). Tube diameters to calculate beam properties from were measured about mid length. Wall
tube thickness was assumed to be 1/16". The real tubes are not round, they are not constant cross section, nor do they have the wall thickness I've assume. I did use aluminum for material
properties which I assume is very accurate.

Below I've taken a picture from the Cruzbike website and added my node location and coordinate system to it. The restraints in x and y direction were applied at the top of the
head set. Restraints in the x,y, and z direction were applied at the bottom of the headset. Finally a rotation constraint was applied at the top of the steering tube. Results for the applied
loads are shown below the graphic.

My intention is to make each tube each tube or set of tubes stiffer one at a time by changing the material properties to steel. We'll get to see which tube or set a tubes makes the most difference
to the deflection. I'm going to guess the forks will make the most difference. Any bets?


Displacements X,Y, Z in inches
Rotations X,Y, Z radians

Node X Y Z rotation X Y Z

1 +1.8806293E-03 -2.4869306E-03 -3.5207125E-06 +2.2608460E-04 +1.7096630E-04 +0.0000000E+00
2 +0.0000000E+00 +0.0000000E+00 -3.5207125E-06 +2.2608460E-04 +1.7096630E-04 +2.6640102E-02
3 +0.0000000E+00 +0.0000000E+00 +0.0000000E+00 -2.0165434E-04 -1.2196175E-03 +2.8596927E-02
4 +2.3011450E-02 +3.5459749E-02 +8.8160966E-03 -1.2439640E-03 -9.8012703E-03 +3.8887698E-02
5 +2.5921476E-01 +3.5471106E-02 +1.7602821E-02 -1.4179566E-03 -1.0258621E-02 +3.9235074E-02
6 +9.5315019E-02 +7.1912094E-01 +1.2535173E-01 +3.1552725E-03 -6.8141357E-03 +4.6293942E-02

The calculated lateral displacement at the bottom bracket (Node 6 Y direction)

 

[billyk]

I'm going to guess the most improvement will occur between Node 2 and Node 6 (part of the flex is in the joint at 2, not included in this analysis).

The reason I think this has two elements:

1) When John Tolhurst designed the Silvio and Vendetta, he stiffened this connection by connecting the BB (6) directly to the handlebars (1) with a beefy tube. I'm guessing he thought and experimented quite a bit to decide that that was the weakest link, and he knows his bicycles.

2) In my experiments, the flex was lessened by stiffening the top (extension) tube. Note that this assembly is not a single tube but has the extension tube fitting inside the boom, and the clamping is only in the top 2-3 inches. The lower part of the boom widens, so the extension tube is free to move inside it, and it does.

I doubt the fork will make much difference, because it doesn't take any twisting strain when the handlebars are pulled on one side and the same-side pedal is pushed. A strong moment is applied to twist the front end around the z-axis. You would like all the force to be in the x-direction, but some of it gets diverted to y. That is the situation where rider effort is wasted by bending metal instead of directing force to the drive train.

Note that on the Silvio and Vendetta, the joint between the boom and the steering tube is not particularly strong, showing that JT knew that it didn't take much strain. Thus we can infer that the forks don't take much strain, either.

If I was building the Quest (assuming the same geometry), I would make the extension tube beefier, and the boom have a constant inside diameter so the twisting force on the extension tube was resisted over its whole length, not just the clamp at the top.

Billy K

 

[MrSteve]

Your work leaves me in the dust, Billy K.

Here's some numberless feedback, for your consideration:

For more aerodynamic advantage, I built a taller chain stay which raised the bottom bracket.
This also was more comfortable, which was a pleasant surprise.

Anyway, the tall wooden chain stay was as stiff as an al-dente noodle.
Over the years, I have added a cross-brace, many layers of fiberglass reinforced epoxy
and even a layer of aluminum flashing laminated around the legs.
Now, my wood-cored chain stay is extremely stiff... which moved the twisting
forces into other, weaker parts of the front frame.

With the front brake locking the front wheel and with the bike held upright on the trainer,
I can try my best to test the front end.
Now, the welded area between the front derailleur on the TFT and the bottom bracket shell will twist a little
under pressure;
the tabs on the bottom of the chain stay (Nodes 4 and 5) will rock a little,
and the steerer tube/front fork assembly will torque a little.
I'm pretty sure that the drop bars I'm using also warp a little... but I don't see it.
So, when one part is stiffer than it's neighbours, the force is transmitted to the weaker parts or areas, on my bike.

My observation of my modified Sofrider V1 -compared to modern Cruzbikes- under load tell me that
the wide TFT welded to the bottom bracket shells on the Vendetta/Silvio bikes minimises the BB twisting under load;
the shorter steerer tube/forks, combined with the TFT tying the BB and handlebars together directly, both minimize twisting under load,
and the shorter chain stays minimize rocking under load.


Finally, the front suspension on my bike does what it was designed to do.
It soaks up some energy.

Kudos!

-Steve

 

[billyk]

Wooden chainstay!!! I would like to see a photo of that! What kind of wood?

MrSteve is right that stiffening one part shifts the flex to another. This is another reason why JT's Silvio/Vendetta design is so good: he simplified the connection between the handlebars and the BB to one simple, strong, tube. With this direct connection, whatever is going on in the rest of the front triangle doesn't matter because there are no other paths to flex.

Billy K

 

[cllsjd]

Per my analysis I was wrong to think changing the fork material would imporve things. The one best thing you could do, based on the analysis, is to make the "steering tube" stiffer. The "steering tube" is my name for the vertical tube that runs from the headset up to the handlebar stem. I made it stiffer by changing the material properties from aluminum to steel. That would make it about 3 times stiffer. Making the tube from the head set out to the bottom bracket stiffer came up a distance second. Of course on the Silvio and Vendetta the steering tube has been removed. So in my next verison of the finite element model I'll pretend I have a Silvio and see what the numbers say.

Here be the results for now.

 

[mzweili]

Nice analysis. Interested to see what will be the results for the Silvio.

 

[cllsjd]

New disclaimer! I don't own a Silvio, I've never ridden a Silvio, nor do I have access to a Silvio.

I made my finite element model for my Quest a bit more Silvio like by moving node 1 to x =-3 y=0 and z=2.270. This puts node 1 inline with node 2 and node 6 about 3 inches behind node 2. In another words I extended the tube from the bottom bracket back behind the headset by 3 inches. I changed the constraint for node 1 from no rotation about the z axis to no displacement in the y direction. This dropped the displacement at bottom bracket from .72 to .28 inches. This roughly cut the displacement of the bottom bracket to about 1/3 of what it was. At node 2 there is about a 200 lb force in the x direction and 250 in the y direction.

 

[Rampa]

A couple people here have moded their Quests that way. Making the upper tube of the sliding boom the bigger one would also help, as that's where that tube's pivot is. That's why the newer S and V have the fatter tube at the upper end.

 

[billyk]

I don't quite get "node 6 about 3 inches behind node 2", but nevermind.

What the "Silvio-like" model has done is to remove twisting of the vertical steering tube, without changing the strength of the extension tube/boom, right?

That is consistent with what Chuck's model says: the weak link is twisting the steering tube. I am skeptical. Equal-size tubes (as the steering and extension tubes are) seem far easier to bend than to twist. I believe that between my foot and hand I can bend the long extension tube/boom assembly an inch or so, but I strongly doubt I could twist the short steering tube 2 or 3 degrees (the equivalent).

I think the difference between the Quest and the real Silvio is the strength of the big boom. Geometry helps, as does removing the joint at node 2, but I'd bet the Q would be a whole lot stiffer if the extension/boom was stronger.

I'd think about modding my Q to strengthen it in this way, but it would certainly make the front end heavier.

Billy K

 

[cllsjd]

Yes, I removed the steering tube from the model. I did nothing to the extension tube / boom.

Billy, it is good that you're a bit skeptical. However, sometimes the engineering is right even when we don't like the answer. The equation for twisting in a tube is

θdegrees ≈ 584 L T / (G (D4- d4) from the engineer's toolbox

L = 11 inches
T = 550 in-lbs
G = 3800000 psi
D = 1.125 in
d = 1 in

That works out to be 1.545 degrees where my finite element model predicted 1.53 degrees. As an engineer I'd accept those answers any day of the week. So my model is right the steering tube twist that much regardless of what you want the answer to be. However, do your own search on torsional deflection of a shaft to find the equation and work the math yourself to make sure that I did it right.

Also remember this much deflection took place under a very high load that I seriously doubt I could generate while riding. Is it really worth trying to modify the bike to decrease the small deflection a more reasonable load would produce? One easy thing to do would be to move the handle bar stem as far down the steering shaft as you can. That would reduce L in the equation and reduce the angular deflection.

Also don't think that the bottom bracket of the Silvio or the Vendatta don't deflect under load, because they do.