Case study: Operations research needs physicists

Operations researcher Stephen Coulson explains how a physicist's skills are valued in a varied and interdisciplinary field that has expanded far beyond its defence origins

How many people do you need to crew a ship? How do you measure the performance of a canoe? And how radioactive is a human being? Questions like these come up all the time in my work as an operations researcher – in fact, I once had to investigate all three of them in just one day.

At its heart, operations research (OR) is about applying scientific techniques of measurement, experiments, trials and analysis of evidence to construct models to predict future behaviour. The term "operations research" was first used in the Second World War to describe the work of scientists, mathematicians and engineers who studied the military and made recommendations for improving its performance. In many cases, operations researchers working on the Allied side gave advice that directly saved lives and helped to shorten the war. For example, an OR team led by the British physicist Patrick Blackett (and embedded with the Royal Air Force's Coastal Command) realized that Allied depth charges were set to explode too deep under the sea to do German U-boats much damage. By calculating the optimum depth for the fuses, Blackett's team was able to improve the kill rate of U-boats threatening British shipping.

Following the discipline's wartime success, OR departments were created in many industries, as well as within government bodies in both the UK and the US. Today OR – which is also sometimes known as operations analysis, or OA – offers many exciting opportunities and challenges for physicists across a wide range of businesses, including health care and energy as well as defence.

A whole-system approach
A typical problem for operations researchers is to understand what resources – people or equipment – are required to perform a specific task. The details of the task can vary: it might involve working out the number of tills needed at a supermarket, the number of beds for an intensive care ward or the number of mine-sweepers required to clear a sea lane. For all of these "balance of investment" problems, as they are known in the trade, the aim is to find the optimum number of resources required to do the job – and also to identify the point at which adding further resources will produce either no improvement or a reduction in performance.

Solving OA problems often means thinking about a task not just in terms of the technical equipment involved but also how people operate equipment and the operating environment. As an example, an OA team was recently asked to identify factors that affect the performance of aircraft taking off from large airports such as London Heathrow. The goal of this investigation was to determine whether it would be possible to reduce the time between aircraft taking off without compromising safety. A number of large, commercial airports in the UK and Europe were identified as suitable subjects to collect data on aircraft take-off. The OA team then collated and analysed information on take-off times, size of aircraft and pilots' notes from occasions when they encountered problems on take-off.

What the team found was that, for the vast majority of take-offs, aircraft were unaffected by turbulence from the previous take-off of the previous aircraft. Hence, it should indeed be possible to reduce take-off times. However, they also noted that in some cases, turbulence appeared to last much longer than expected. Further analysis showed that long-lasting turbulence did not appear to be correlated with aircraft size or engine configuration. Instead, it seemed to be weather-related. Most of the airports examined in the study collected highly accurate data on meteorological conditions during take-offs, such as wind speed, air pressure and temperature; using these data, it was possible to identify specific weather conditions that delayed the dispersal of turbulent air. When these conditions occurred, air traffic controllers at the airports were advised to increase the interval between take-off times.

This example illustrates the importance within OR of thinking about problems at the level of systems, rather than individual components. Often, this involves creating models to measure and compare the performance of many different parts of a system, from sensors to complex equipment (such as life-support machines or 3D printers) and even the training requirements of the people operating the system. This kind of thinking requires knowledge of many different areas of science and engineering, which is why OR teams often include a wide range of scientists and specialists – including some you might not immediately think of, such as behavioural psychologists and historians, as well as mathematicians, natural scientists and computer programmers.

Getting into the field
I first came across OR while serving in the Royal Air Force, where I met a number of operations researchers who were embedded within military headquarters. Before joining the RAF, I had completed my PhD in maths and theoretical astronomy, and I found I really enjoyed talking to the OR teams and learning how they could help with mission planning. After I left the military, a career in OR was an obvious choice, as it combined both my military experience and my physics background. For me, being part of a highly multidisciplinary environment is one of the main attractions of working in OR, as discussing problems with a diverse range of specialists often leads to some highly original ways to solve problems.

One of the main changes in OR over recent years is that large industrial firms and government bodies have reduced the size of their in-house research departments. This means that the majority of operations researchers tend to work for small and medium-sized enterprises (SMEs), which are then employed to solve OR problems for industry and government. Many of these SMEs are now looking to recruit recent graduates and postgraduates to train as operations researchers.

The good news for physics students is that the blend of problem solving and research skills acquired during an undergraduate physics course makes them attractive candidates for OR teams. Being highly numerate and experienced in both theoretical and experimental work, physics graduates often already have many of the skills required to become good operations researchers. If variety, working closely with customers and looking for original ways to tackle real world problems appeals to you, my advice is that a career in OR could be for you.


Advice for people interested in an operations-research career

Students in the first two years of their undergraduate degree programme
Many operations research (OR) firms offer summer placements or placements for those on a "sandwich" degree course that includes a year working in industry. If you think you might be interested in a career in OR, it is worth contacting firms specializing in this area to discuss the work they do and any future opportunities they have.

Final-year students
A postgraduate qualification is a useful, but not essential, means of gaining further experience in the skills required to do OR. A number of universities offer specialist MSc courses and PhD programmes in OR, usually through their mathematics departments. Postgraduate OR courses nearly always include strong collaboration with either industry or specialist OR firms.

Looking for a career change
Many OR practitioners have come into the field following a career in industries directly supported by OR such as logistics, healthcare or defence. As well as bringing their technical skills they often have good knowledge of how systems operate within their specialist industry. Alternatively, some people find they are already doing OR-type projects within their current employment before deciding to specialize specifically in OR. A colleague of mine spent a number of years working in radar systems before taking a career break to canoe down the river Nile. Following his expedition he was attracted to OR as it gave him the chance to combine his experience in analysing networks with an understanding of human factors.

About the author
Stephen Coulson is head of operations research for RED Scientific, a Hampshire-based company specializing in operations research, systems engineering and scientific consulting, e-mail or@red-scientific.co.uk