By Dr M.C. Duffy
From the August 2006 Newsletter
Railway mechanics grew from the 19th Century analysis of the locomotive and track machine-ensemble, formed by the union of locomotive and track. Improving the fit between engine and track was essential for safe and progressive operation. Analysis of the New Machine stimulated general engineering mechanics. Poussin, Clerk, Reuleaux, Reynolds et al. studied the balancing of locomotive mechanisms in the 19th Century and the phenomenon of fatigue failure was first identified in a locomotive context. In the early 20th Century, the study of shatter cracks in rail heads resulted in new regulations governing the manufacture of rails. Sylvester and Tschebyshev considered the kinematics of mechanisms, including valve gears. O. Reynolds studied balancing of reciprocating parts, and coupling rod failure. Railway studies of materials are reflected in the work of Mohr whose analysis is still used in strength of materials.
In the 1890s, mechanics was supplemented by thermodynamics and quantified analysis of energy flows were carried out in locomotives. The electric power industry used energy analysis which passed into railway engineering after 1900. In the 1880s Wellington published his general theory of railway location, linking engineering and operation to climate, topography, commerce and economics. The influence of this work is still felt. In 1900, the machine-ensemble (traction system) was integrated closely with signalling via interlocking. Systematic investigations into locomotive proportions were conducted by Goss at Purdue, following earlier studies by Borodin in Russia, and Le Chatelier in France. Le Chatelier was a pioneer of rational management, efficient energy use and scientific organization, as were F. W. Taylor, Gilbreth, Gantt and Ford in the 20th Century. Their theories transformed railway engineering management and design. The 20th Century saw American exemplars transform every department of the machine ensemble: traction, permanent way, signalling and control. New devices were in part responsible: track design, rail shape, track circuits, power signalling, automatic signals, mechanized equipment in depots; electrical safety mechanisms; but revolutionary concepts in management and organisation drove much of the technical changes. Electric railways witnessed early attempts to model the technical and economic features of large industrial systems.
The work of Wellington in the USA modelled the railway in its general environment, and the work of Lomonossoff in Russia resulted in a comprehensive theory of Railway Mechanics in the early 20th Century. He advocated research departments dedicated to railways mechanics to frame rational strategies for engineering development. Lomonossoffs theories and experiments showed the difficulty of assigning accurate, measured values to parameters in theoretical expressions - without which comprehensive theories were of little value. Modern testing techniques were developed to overcome this problem. Experiment and experience were insufficient to solve major problems and theoretical analysis was needed. The Russian contribution was considerable. Grinevetsky, Syromyatnikov, Nikolayev, and Shelest developed academic courses on locomotive design between 1900 and the 1960s. Similar academic work was done in Germany, and the USA. Steam locomotive science was developed in France by Chapelon; in the USA by Woodard and Johnson.
The effects on track of out-of-balance forces in steam locomotives was made in the 1940s tests carried out on the Chicago North Western RR. The behaviour of rails, trackbed, and locomotive components were measured using electrical recording apparatus, and high speed cameras. Wheel lift was observed during slipping. A mechanical model of wheel-rail interaction was constructed using rollers and springs, which could represent different rail dimensions, ballast characteristics, axle loads and train speeds. This helped compare theory with observations. Guidelines were derived governing springing, wheel design, balancing and number of cylinders. The tests were repeated under British conditions by the LMS railway in 1941. The LMS set up a research centre in the 1930s after the American model which investigated train resistance, aerodynamics, and vehicle-rail interaction.
Carter pioneered general analysis of rail-vehicle interaction in 1916. Hunting or sustained oscillations due to dynamic instability lacked adequate theoretical interpretation, as did creep and the dynamics of curving. These affected all vehicles on all kinds of track. The growth of oscillations at high speeds, even on straight track, limited very high speed operations. Graphical analysis was often needed to solve equations. Carter's work, though of great importance, was limited in value because the configurations of vehicle body, wheels and bogies, which he considered were inherently unstable. The analysis was developed by Rocard in France. Many railway engineers tolerated hunting as inevitable but the high speed tests of the 1930s in Germany, Great Britain and the USA, drew attention to the lack of understanding. In Great Britain the LMS railway contacted Prof. Inglis at Cambridge University. The Cambridge work, published in 1939, included equations of motion for a two-axle bogie but the 12th order expressions could not be easily solved by the techniques then available.
Advances were made after World War Two by Matsudaira at the Railway Technical Institute, Tokyo who developed the concept of stiffness to stabilize spring-restrained wheelsets at any critical speed. By the 1960s this work enabled the Japanese railways to operate very high speed Shinkansen trains. The problems investigated concerned all forms of train, and in the 1960s British Railways contracted Imperial College to develop adequate analysis. In both the Japanese and British work, dynamicists with a background in aeroelasticity played vital roles (Matsudaira, Wickens). The theoretical analysis was tested by experiment using roller test stands, and instrumented trains and track, greatly aided by the availability of computers to solve expressions which were difficult or impossible to solve by earlier methods. Developing adequate theories of vehicle behaviour on curves was difficult. An intuitive approach by Mackenzie in Britain (1883) was followed by more analytical investigations by Nadal in France (1896), and Boedecker (1887), von Helmholtz (1888), and Uebelacker in Germany (1903). A fundamental contribution was made in Britain by Porter whose studies (1934) on mechanics of a locomotive on curved track became classics. By the late 1960s, fairly complete theories of curving were in existence, which included vehicle form, creep and stability (Boocock, 1969). In 1977, a general theory of curving was formulated by Elkins and Gostling, 1977).
Current research and development involves co-operation between state, manufacturers, universities and railway companies. There is a trend towards international co-operation, encouraged by globalisation of leading manufacturers. The German programme into advanced railway technology begun in 1972 is one example. Government funding varies between 50% and 100% depending on the project. Federal funding in the USA likewise varies with the project. The German programme covered vehicle and rail interaction, vehicle design (including aerodynamics and control), track and trackbed design and behaviour, control technology, power technology including pantograph-contact wire interaction, and environmental matters. The ME-DYNA programme was structured for non-linear simulation of vehicle track interaction. The Munchen-Friedmann test facility was constructed 1977-1980 to investigate four-axle vehicles. Wheel sets have been tested to 503 km/h. Techniques for predictive analysis of bodywork stresses, modes of vibration, distortion during all modes of motion were improved. Aerodynamic behaviour of vehicles and pantographs was optimised. Without this analysis, dependent on electronics for its execution, the modern railway, with its enhanced performance, and close integration of traction system, signalling and automatic control would not be possible.
