Friday, 21 February 2014

Walking and working the line, and riding the buses

The late Alec Glaskin was head of O.R. at Pedigree Petfoods in the 1960s and 70s, and was reputed to earn the highest salary of anyone in the U.K. working in commercial O.R..  He insisted that all recruits to his team should spend time "walking and working the line".  In other words, they must have first hand experience of the production process which they were to model.  Many other senior figures in O.R. have taught the same principle.  In the U.S., Gene Woolsey was a leading exponent of the necessity of knowing how people really did things.  Experience of the world being modelled is essential for the modeller.  More recently I came across a study of the expansion of a supermarket chain into convenience stores; the young O.R. worker assigned to studying the distribution spent several months travelling in supermarket lorries to observe the process.  As a result of her observations and some astute modelling, the company changed its procedures, because it was clear that distribution models for out-of-town supermarkets did not apply to the smaller shops - obvious, but her work specified how they did not apply.

This week the Labour peer, Lord Adonis, who is standing as a candidate for the post of Mayor of London in a couple of years time, is spending five days travelling London's public transport.  He is riding the buses - walking the line, so to speak.  He has written about his observations, things that he has seen which he might have learnt from other people, such as his political advisors, but which come much better by first-hand experience.  That's the parallel with O.R. experience at the sharp end - sorry, that expression is a cliché, but it is a cliché that works to describe what he is doing.  He notes the impact of good transport on the economic well-being of an area.  And he decries the imposition of a charge by the bus company (Transport for London) for bringing bus routes into an area of new housing, a charge which will add nearly £3000 to the cost of each new house there, if the houses are built.  I hope that it wasn't an O.R. person who suggested that charge!

There aren't many similarities between politicians and O.R. scientists, but here is one - the importance of experience of the real world, not the synthetic one of models and paperwork.

Sunday, 16 February 2014

Shipping containers

The London Gateway is a newly opened container port on the River Thames, close to London and the transport infrastructure of south-east England.  It has been designed to handle some of the largest container ships in the world today.  Our Saturday newspaper had a long article about this site.  Having read the article, there was one fact that could be used by Google as the basis of an interview question: "How many movements of shipping containers happen each year in the world?"  Tina and I thought about how we would approach the guesstimation of the answer.  We know that many millions of cars are moved each year in shipping containers, so transport of vehicles accounts for somewhere between 5 and 50 million container movements, and vehicles must be one of the larger uses of containers ... so we reckoned that we would estimate between 100 million and 1000 million container movements.  The answer in the article is 500 million.

At one O.R. conference that I attended, there was a speaker who described how the growth of container shipments and the ports that handle them had revolutionised international trade.  But, he argued, the revolution depended on the simultaneous growth of computer power, and the development of algorithms for managing those containers and their contents.  Which is why the lecture was given to O.R. people.  Efficient container management requires forecasting of demand for goods, months ahead.  It requires supply chain management.  It requires fleet management for the ships and dockside equipment.  It requires routing algorithms for loading and unloading the ships.  It requires personnel forecasting and management for the fleets and ports.  Many of the containers are used in "Just in Time" supply or manufacture.  The list of areas of O.R. expertise goes on and on.  To make the article appeal to a non-specialist readership, the news article was headed "It's like a giant Tetris" linking the colourful Tetris blocks to the colourful 40ft containers.  Maybe experience with devising rules for beating the computer at Tetris can be used to suggest heuristics for moving these larger coloured blocks?

Sadly, despite the many contributions of O.R. to international container shipping, the term did not appear.  We really are the "Hidden science"!

How many people are coming to this event?

The Institute of Mathematics and its Applications is a UK-based professional society for mathematicians in industry and academic life.  Many of its members are also members of the OR Society, but the IMA extends to mathematics in many other areas, including engineering and computing.  The IMA has a high profile in education, and produces publicity material about interesting and unusual aspects of mathematical models.  The series is called "Mathematics Matters".  One of the most recent items is about crowd modelling, "Following the crowd" which draws on work by Professor Keith Still on managing crowds for safety and security. 

"Following the crowd" highlights the problem of deciding how many people are coming to an event, and how to control access.  The example taken is the attendance at London's New Year fireworks display.  It is free, so there are no advance bookings to work from.  The safety concern is when to close access to the viewing areas; leave it too late, and either people are herded into a dense mass, or the crowd is forced to turn round against the flow ("Those behind cried 'Forward', Those before cried 'Back' ").  The outline of the method is to take arrival rates at regular intervals, fit these to a model of the pattern of arrival rates, forecast attendance from this model (or at least forecast the arrivals for a few future time intervals) and close off access to feeder channels when the viewing areas hold about 80% of their capacity.  The feeders will supply the remaining 20%. 

Keith Still's website has numerous other stories of models of crowds, and of what happens when there is no organisation, or inadequate organisation.  It is instructive to see how this area of modelling brings together forecasting, queue models and network flows.  It extends to the design of feeder channels - as used in theme and adventure parks as well. 

Monday, 3 February 2014

Maintaining hedges

Two years ago I blogged about regular maintenance and included a brief discussion of how often hedges on farms and on the edge of roads should be trimmed.  I had taken this example from some guidance in a leaflet from our Devon County Council.
Contrasting hedges on the two sides of the road

Earlier this year, I found a further objective in the list of conflicting objectives that appeared then.  This is one which can be expressed as a constraint.  (Treating objectives as constraints is one of the tools of multi-criteria decision-making.)  Some hedgerow trees produce their flowers and fruit on second-year growth of the plant.  So if the hedge is trimmed once a year, then the plants never produce blossom or fruit, which is  a disservice to wildlife.  So such hedges should not be pruned more often than every two years, but three year cycles are obviously possible.  So, to help wildlife, there is a constraint if the hedge contains such species.

The same source of information also commented on the need to vary the method of cutting hedges back; if mechanical flails are the only tools used, then the young growth on the hedge is cut back more drastically than the established growth, and the base of the hedge can be opened as the hedge trees grow taller without fresh shoots at the base. 

So this is a supplement to the earlier blog, a reminder (to me as well) that extra information for an O.R. study can be found and be used, even after many years!