predicting real world axle failure and predicting real...
TRANSCRIPT
Introduction:
Axle fatigue overview
Roderick A Smith
Co-Chair TC24Future Rail Research Centre, Imperial College London
Predicting real world axle failure and
Seminar: 3-4 March 2011
Predicting real world axle failure and reliability
The Versailles accident of8 May 1842.
Broken axle - derailment of engine – fire in wooden carriages – doors locked –approx. 70 killed
An internationally reported sensation.
First railway accident to cause major lossof life
The origins of study of the fatigue problem generally and axles in particular
Versailles, Sunday 8 May 1842
Catalyst of much early work on what became known as “fatigue”
Interesting reports of discussions in Proceedings
I Mech E, I C E.
In 1804 Richard Trevithick demonstrated the first working rail locomotive.
In May 2004, Railfest in York celebrated this notable bicentenary.
Saturday 29 May 2004
Established the concept of a fatigue limiting stress range
Franz Reuleaux, 1829-1904
German entrepreneur, design engineer and scientist
Method of calculation of bending stress in axles given in the “Constructer”
( Chapter viii, Axles)
First edition 1861.
Longstanding problem:Fatigue• The science is (rather) well known• The longer the crack, the easier it is to predict its growth• Initiation is very difficult to quantify and is controlled by local factors
(defect, corrosion, fretting, notch etc.)• Material properties are generally well known and quality has
improved• Quantification of laboratory tests is generally easy because the
loads (stresses) are well defined and controlled• Similitude between laboratory and real world often difficult because
in the real world loads are often ill defined and generally of variable amplitude
Inspection• NDT to reliably identify and size cracks has long been an Achilles
Heel
Problems for railway axles:• Load measurements need to be converted to stresses in
the critical location• Not many axles fail!• Light weighting is becoming increasingly important to
save energy (minor role) and to reduce dynamic loads (major role)
• Inspections need to be managed to minimise possible damage
• Crack initiation dominated by relatively few high stresses, the full spectrum playing an increasing part as the crack length increases
• Life between a reliably detectable crack size and final failure is likely to be short!
m
Most “design” is based on some form of empiricism which works over a limited range of conditions: trouble occurs when extrapolation takes us out of the “comfort zone”