Re: Forces on Cranks
- From: Jobst Brandt <jbrandt@xxxxxxxxx>
- Date: 30 Apr 2010 21:23:08 GMT
Michael Press wrote:
Thank you for the write up, Frank.
The designers have the tools, but do they type in _all_ the
forces induced by the pedal?
The interested reader might consult Landau & Lifschitz,
Theory_of_Elasticity, pp 68-75. They derive the theory of
torsion for rods, and work out the torsional rigidity for
rod with cross section circle with radius R:
These days, any company can afford to buy a finite element
analysis program that will run on a personal computer. In fact,
there are many packages that are free and open source. They take
knowledge and some skill to use correctly, but they're far more
reliable than the sort of simple approximations allowed by such
Many, if not most, commercial packages will let you produce a
solid model of a proposed part, then do stress analysis. Once
the model is built, mesh generation is nearly (maybe not
perfectly) automatic. Once the mesh is generated, you can run
analysis of any load you choose. If you don't like the resulting
stress levels, you can redesign before you ever cut metal. It's
far less expensive than the old days, when every stage of the
prototyping required metalworking time plus physical testing.
Once the model is verified, it can (with proper software) be used
to generate CNC code to machine positive prototypes, or negative
dies for casting or forging. If preferred, rapid prototypes can
also be produced in plastic or even metal. Other, more
specialized FEA software can be used to help ensure the proper
flow of metal in dies and molds.
I know one engineer whose company does all the above for low-
production replacement parts for humans - i.e. artificial hips
and the like. I assume companies like Shimano and Campagnolo use
such processes to design and manufacture their products. It's
likely that small companies producing boutique CNC parts don't go
this route, but in those cases, I think you don't get what you
Again, these processes have now replaced a lot of approximations
via simple stress formulas, at least for production
parts. (They're not necessary for simple structural parts in
non-weight-critical applications.) They also replace Jutish
attempts at visualization - although any competent design
engineer will cut out a lot of nonsensical Jutish speculation
just by his understanding of stress distributions.
Does not answer the question. Do they type in the bending moment
on the crank out of the plane of the chain wheel? Do they read
In the packages I've used, if I were designing a crank, I'd be
looking at a mesh model of the crank. It would be critical to
properly represent the boundary conditions (say, attachment to the
crank axle) and the loads; otherwise, results would be erroneous.
I could either apply a force and a couple at the location of the
pedal threads, or I could model the pedal axle as an extension to
the crank and apply a force at the appropriate place on that pedal
axle. Either strategy could be quickly repeated for force and
couple in any direction, once the mesh model was created. That
would allow study of the complete 360 degrees of crank rotation.
"Reading the answer" would consist partly of looking at the color
coded stress levels shown on the model, and reading information out
of resulting tables. You also get to see deflections, if you're
interested. (Or if you're a person who Jutishly confuses stiffness
But realistically, those folks wouldn't get far enough to use the
Here's a site that has some screenshots of their software:
Again, I'd be shocked if Shimano, Campagnolo, and even Chinese or
Taiwanese manufacturers, weren't using the processes I described.
The recent hollow crank from Shimano is a jungle of stress risers.
New Al cranks continue put all the stiffness in the direction it is
least needed: in the plane of the chain wheel.
That is the proof that they don't and are incapable of properly
analyzing the stress on cranks, as I have said for years now. They
also don't understand why they use left hand threads or they would
stop doing that and also not load threads with dynamic radial loads.
No FEA will help them understand this because it is a dynamics and
tribological problem, not a stress problem.
- Prev by Date: Re: The engineering ignorance of Frank Krygowski
- Next by Date: Re: Forces on Cranks
- Previous by thread: Re: The engineering ignorance of Frank Krygowski
- Next by thread: Re: Forces on Cranks