A Train of Thought
April 12th 2013
Earlier this year I travelled from London to Glasgow for a meeting. That gave me the opportunity to ride on a railway train for the first time in Britain since the era of British Rail. My memories are of a system that was starved for funds with much neglect and decay. Returning by rail to my hometown on one trip was a shock. The elegant Victorian station had been savagely removed, and a corrugated iron shed had been put in its place, losing a few hundred yards of track.
However the journey to Glasgow drove away all those memories – it was a brilliant ride! The journey only took four hours and it was comfortable, quiet and very relaxing. It was tempting to sit back and thoroughly enjoy the experience, instead of working as planned.
One of the main improvements was the ride smoothness at around 125 mph. I had to do some ‘engineering’ experiments to detect the vibration (I will leave you to guess what those were and perhaps get some feedback on ideas, in the spirit of exploding dough and carrots). I walked along the carriages without the violent swaying and rocking that I remember well. I have since read an article explaining that it is a combination of track straightening and banking, together with carriages that can tilt into the curves. The combined bank angle was impressive; with the surrounding countryside rolling about the carriage axis quite happily.
The evolution of that particular railway route is remarkable. The first stub of the route was created in the 1830s as a junction from the Liverpool to Manchester Railway. Over the next few years other parts were added by emerging railway companies. However, there was originally no overall plan to have a continuous line running from London to Glasgow operated by a single company. The route taken today is as a result of a piecemeal approach with many compromises along the way.
We can think of this in terms of a structural optimization problem. The analogy is a design which has evolved to meet a series of disconnected local objectives with corresponding constraints. The design is then rethought and re-optimized with unified objectives, however many of the constraints from the old specification still apply. The technology jump could well be represented by a move from metallic materials to composite materials introducing a fresh set of design variables.
Clearly the re-optimized design suffers from not starting with a clean sheet. Just like the railway line, many assumptions and compromises are locked into the configuration. Some of these will be sound engineering judgement and will benefit the structure; others will keep the structure from becoming a ‘true’ optimum. I think many designs evolve along these lines in practice. It is difficult to quantify good judgement and to keep it in the design. Methods such as Genetic Algorithms and Neural Networks can present us with configurations where we can perhaps identify good engineering. Posing an optimization problem as multi-objective is powerful as it keeps us in the decision loop, assessing best candidates.
The UK government is planning new high-speed rail routes to link London and the North of Britain; starting with a clean sheet after nearly 200 years. Tight curves which drive a lot of the current technology are eliminated. The number of stations is low; somebody must have a multi-objective study looking at speed, station revenue and probably voter appeal!
Luckily for us on the structural engineering side, politics is one of the variables that we can leave out. Although having written that down I realize it is probably a very naïve statement!
Until next time,
P.S. the on-line e-learning course on Optimization starts on April 16th.