This article is from the Bicycles FAQ, by Mike Iglesias with numerous contributions by others.
From: Jobst Brandt <firstname.lastname@example.org>
Date: Wed, 03 Jan 2001 16:50:20 PST
How to determine whether a frame is straight after a crash and what
can be done about it.
First is visual, especially for head-on collisions on a standard steel
frame, on which top and down tubes generally bend at the end of their
butted section, about 50-100mm from the head tube. This usually
causes cracks in the paint and can be detected by laying a straight
edge on the down tube. Next, sight down the fork to determine if the
fork blades are straight in the fore and aft plane, and whether their
upper straight portion is parallel to head tube. Bicycles with
straight blade forks (with angled crown) make the latter impossible.
Another simple test is to ride no-hands and see whether the bicycle
rides straight. This will show whether the fork is laterally correct.
Determining whether the "rear triangle" is displaced requires
measurement. The rear triangle, actually a tetrahedron (four sided
figure with six edges), is not easily bent except by side force on the
BB. Tubes bent by a force at midspan are self evident by no longer
being straight. Bicycles with curved stays are on their own here,
having no credible reason for their curvature, which becomes apparent
when trying to determine whether they are "straight."
Rear triangle displacement is measured by stretching a string from one
dropout over the head tube back to the same place on the opposite
dropout. The distance between string and seat tube should be
identical for both sides. Also, because the two sides of a frame are
seldom identically strong, dropout spacing will most likely not be
correct, one side having yielded differently than the other.
Such lateral displacements can be manually corrected by laying the
frame on its side, placing the foot on the inside of the lower
chainstay at the BB and pulling the dropout of the upper side toward
the correct position. Monitor position change by measuring dropout
spacing. After advancing a few millimeters, put the foot on top of
the upper chainstay at the BB and pull the lower dropout until the
spacing is correct and repeat the sting measurement.
Laterally correcting a front fork is done similarly while monitoring
dropout spacing. Here the critical test is whether the bicycle rides
no-hands straight, which is relatively easy considering that the only
the wheel need be removed to perform the bend. Otherwise, sighting
down the head tube onto a dummy axle with a centerline on it can help
determine whether the fork is "on axis." Forks are best straightened
with fixturing but can be done without.
For steel frames, these operations pose no problem if the distortion
is within limits that do not peel off paint. Frames with oversized
tubes generally make their fatal bends self evident by wrinkling as do
downtubes of standard steel frames in head-on collisions.