Index Home About Blog
From: Mark Bulgier <>
Subject: Re: Curved Seat Stays - smoke and mirrors?
Date: Fri, 24 Oct 1997 12:30:04 -0700

Bruce Frech wrote:
> What is the major stresses on seat stays
> that determine how strong & stiff they need to be?
> My guess is the bending due to braking is the major factor with
> holding the seat tube away from the rear wheel a distant second and
> torsional forces even less.
> If the braking force is what drives the design then the minor curves
> in the Lightspeed road frames add a little style without compromising
> the design constraints.

Bike frame design has advanced for over a century by *almost* random
mutations (well hopefully some of the designers had *some* clue!).
A recurring theme is making a part lighter until the death toll becomes
prohibitive, then adding a little material where the breakage occurs.

In seatstays, it's hard to make 'em too light for the stresses they see
on the road because those stresses are so slight, that a stay that
would fatigue early for many riders would be too weak to survive the
framebuilding process.  There's a practical limit to how thin a tube
can be mitered and welded or brazed, or cold set for alignment without
kinking.  The lightest stays require very skilled labor to make joints
without damaging the tube.  The rare cracks I've seen in the joint at
the seat cluster, or more rarely at the dropout, were almost always
attributable to imperfect welding or brazing.  Ritchey and Lightspeed
are two of the best in the business, so I wouldn't worry.

The upshot is seatstays hardly ever break.  Unfortunately this means
I can't learn about the stresses they see by looking at lots of
fatigue cracks -- their location, direction, shape and texture -- like
I can for most every other tube on the bike.  So I'm left with pure
conjecture, and mine runs along the lines of yours except with the
clarification that the stays aren't really sized to handle the riding
stresses but to survive the "birth trauma" in the frameshop.

S-stays are a factor, though maybe not a big one, in the harmonic
frequency of the bike (or more to the point, of the bike/rider
_system_) but this is notoriously difficult to analyze.  I have not
seen any simple correlation between extra light s-stays and high speed
shimmy, for instance, though it could be a factor for some.

Pannier racks and canti brakes add a rigidity constraint, but if the
stays are sized up as they usually are for racks or cantis, then
they're even more overbuilt WRT strength and thus even less likely to

It's interesting to note that the lightest aluminum road s-stays from
Easton are hardly any bigger than steel stays, though their metal is
roughly half the strength and one-third the stiffness of typical bike
steel.  I haven't ridden one of those bikes; do they suffer any ill
effect from such light s-stays?  I don't expect they do.  With poorer
fatigue characteristics than good steel, they can't flex much or they
wouldn't last long.

Mark Bulgier, Seattle

Subject: Re: tensions in a bike frame
Message-ID: <EDZwa.14579$>
Date: Fri, 16 May 2003 04:17:40 GMT

Phil Brown writes:

>> Hi, did any of you try to calculate the tensions in a bike frame,
>> for example using a finite element software?  any web site?

> Not much tension in a bike frame.  Most loads are compressive.

Having had involuntary research on this, I can say that the seat tube
is in compression while sitting and tension while standing.  I rode
home sitting when the seat tube broke at the BB.  Chainstays are in
tension except the right one when pedaling hard.  The larger diameter
tubes also experience bending and torsion, something the "rear
triangle" tubes do not (that's why they are thin.)  Braking forces
(bending) at the rear brake bridge are insignificantly small.

Analyzing stresses in a bicycle frame is difficult because the human
input is hard to model to a large degree and valid assumptions are
difficult to make.  A good method would be to instrument a bicycle
ridden by a versatile rider to record loadings in frame elements of
interest during various modes of use.  Then an analysis for peak
stresses could be made using these inputs.  I believe this hasn't been
done to any significant degree or we would have seen (promotional)
reports of it.

Jobst Brandt
Palo Alto CA

Index Home About Blog