Podcast | Operations Research
Can Operations Research have applications in everyday life?
Hear Marc West discuss the scientific field of Operations Research.
Transcript
[Music plays and - fades]
Mish Kardachi: Hi, I’m Mish Kardachi with another podcast from Defence Science and Technology.
Mish Kardachi: In today’s podcast we hear from Defence analyst Marc West who explores the fascinating world of Operations Research.
Mish Kardachi: Not many of us would consider there to be a connection between the complex field Operations Research and areas of everyday life such as sport. Marc West will have you think differently.
Marc works at the Defence Science and Technology Group as an analyst, largely in the field of modelling and simulation. He provides advice to the Australian Navy on tactics and requirements. Marc is an experienced science communicator who actively promotes science and technology throughout his local community.
[Music plays - fades]
Marc West: Whenever somebody asks what I do for a living, “Operations Research” usually gets a blank stare. It is not a well-known field of science. But put simply, Operations Research (OR) is the application of science to decision making. We write models to inform better decisions.
While this quick and dirty statement of OR makes a nice sound-bite, what it is in practice is not universally agreed. Some consider OR to be an evolving body of scientific techniques that is applied whenever a new problem comes in. Others consider OR a process, separate from the techniques, that is focused on solving the decision maker’s problem using the scientific method. In the Defence context, DST Group defines OR as “enabling the evaluation and analysis of military problems to provide decision makers with a scientific basis to improve military operations or capability [1].”
History
OR, by definition, has a human making a decision, so if you want to be pedantic about it, the history of OR stretches all the way back through the history of humankind. Back in the 5th Century BC, arguably Sun Tzu first captured the essence of military OR in The Art of War [2], but it is generally agreed that, although particular techniques can be traced back to earlier origins, modern OR became a fully-fledged discipline during World War 2. As the problems being tackled were too complex to be solved by single people working by themselves, multi-disciplinary teams were brought together, the most famous of which was “Blackett’s circus”. Patrick Maynard Stuart Blackett, a future Nobel Laureate and former naval officer, was Director of Naval Operational Research in the British Admiralty from 1942 to 1945 who, along with Evan James Williams, a Welsh experimental physicist, worked on a number of OR problems that would become text book examples of the discipline.
In 1941-1942, British Coastal Command was having difficulty defeating German submarines from the air. Williams observed that most successful attacks against submarines were made while the subs were on the surface, or barely submerged, before they could run away. However the depth-charges being dropped were programmed to explode at a depth of 100 feet, much deeper than the typical depth of an escaping sub. Williams recommended that this distance be changed to 20-25 feet, which resulted in a 400% increase in the destruction of submarines [3, 4].
Another classic WW2 example was the question of how big convoys of merchant ships in the North Atlantic should be to minimise the number being sunk by submarines. Statistics suggested that losses were proportionally smaller for larger convoys, but the analysts were not convinced, so they set about trying to understand the cause of this correlation. One vital factor was the submarine’s ability to detect the convoy, and it was shown that it a convoy, due to the tight packing of the ships, was almost as difficult to detect as ships on their own. And it was certainly much harder to find a convoy, irrespective of its size, than it was to find a ship if all the ships in the convoy had scattered and headed out on their own. If the submarines were able to find the convoy, a large convoy was proportionally easier to defend than a small one, and then if a submarine did successfully breach the escort screen, whilst incredibly difficult to defend against, the submarine would exhaust its torpedo supply quickly, limiting the amount of damage it could cause. Blackett so convinced himself of their analysis that he said “if I were to send my children across the Atlantic during the height of the U-boat attacks I would have sent them by a big rather than by a small convoy”. The Allied change of tactics from small to large convoys was adopted in early 1943, with substantially reduced shipping losses paving the way for the D-Day invasion of 1944 [5].
Defence Operations Research
OR is one of the core capabilities of DST Group, but DST wasn’t always DST. Before DST Group was formalised as a nationwide entity, it existed as a disparate set of organisations working for the armed services, as research arms of government departments, or as part of the Council for Scientific and Industrial Research (CSIR), the precursor to the current CSIRO. In Sydney, OR was conducted under banner of the Royal Australian Navy Experimental Laboratory (RANEL), which formed in 1956 at Rushcutters Bay. RANEL changed its name to the Royal Australian Navy Research Laboratory (RANRL) in 1969, and in 1975 was amalgamated with several research organisations to form the Defence Science and Technology Organisation (DSTO) [6]. Similar developments and movements created Aeronautical and Land OR capability in Victoria and South Australia, and Central Studies Establishment in Canberra has now evolved to study Joint and complex OR [7, 8].
Maritime OR in the early days examined some classic WW2 problems such as anti-submarine warfare escort requirements, and the defence of small convoys. Since then, OR capability has been brought to bear on many needs and requirements decisions for all sorts of ships, including Mine-Hunters, Destroyers (and their replacements), Frigates (and their replacements), Patrol Boats, and recently the ginormous Amphibious LHDs. And yes, convoys! [9]
Sport
With ever increasing computational power, OR is finding more application in everyday life, with one of the more interesting areas being sport. “Moneyball: The Art of Winning an Unfair Game” [10] about baseball team the Oakland A’s use of data and models, published in 2004, has become synonymous with OR in sport, but in the 13 years since its publication, big data, high-speed internet and wearable tech has allowed us to probe even further, with most professional teams in all codes of sport employing data analysts in an attempt to find an edge over their competitors. But it’s one thing to have the data; it’s another to turn it into something useful, and this is where the lessons of Moneyball are truly learnt. What are the important parameters to be capturing? What are the variables that actually contribute to your team’s chances of winning? Similarly, what are your aims? Is a win this year what you are after, or do you want to win the comp next year, or a World Cup in 3 or 7 or 11 years? Or is your sporting team a business where finals appearances each year, perhaps at the cost of winning the grand final, might be your aim in order to keep the business running?
Not only are the analysts within each team structure interested in this, but so are the fans. If you work in OR and are a fan of sport, you've probably at some point developed your own ranking systems. I know I did. From scribbling down cricket statistics as a kid, to developing my own ridiculous financial market for cricket teams (you can google it!), I've always loved mucking around with sporting data. But more than mucking around on the back of an envelope or in Excel (the modern equivalent), sporting OR is now big business, with sporting ranking systems developed with wide application from betting markets to team development.
Similar concepts exist within military OR. How should you best constitute your force in order or overcome a future adversary? What are the effects you are trying to achieve, and what are the most important factors in achieving them? These are concepts that influence how the military conduct their current operations, and how they will shape themselves in the future. The big difference between the sporting and military contexts of course is that there are plenty of sporting games played and therefore lots of data - this is not the case with the military. One way military OR analysts attempt to overcome a lack of data is through simulation, and with computers becoming ever more powerful, so does our ability to model the world. But all the world’s computing power won’t help you if you don't have well defined questions that you are trying to answer. In fact, it could bias your model. A classic sporting example is the 2009 Ashes cricket series. Australia had 7 of the top 8 run scorers for the series, and the top 3 wicket takers, yet still lost. A cricket model focused on these major statistics would have failed in predicting the result.
Whilst DST isn’t in the business of contracting our services to professional sporting teams (as fun as that might be!), techniques developed in-house can be applied to sport; indeed, given it’s hard to publish in many areas because of security concerns, application to sport is a creative method of seeking external peer review [11].
Gender Balance
The first international conference on OR was organised by the OR societies of the UK, US and France and held in Oxford, England in 1957. It was attended by 250 delegates from 21 countries (with 7 Australians [12]), but a close look at a photo of the participants shows that of the 201 delegates captured by the shot, only 5 were women. Shortly after this conference, the International Federation of Operations Research Societies (IFORS) formed and sponsored the triennial series of international conferences that continues to this day – I attended the Melbourne version a few years back. Whilst the gender balance of subsequent conferences has improved, OR has historically been male-dominated. The IFORS OR Hall of Fame was created in 2003 and currently has 23 inductees – all male. Even though panel members were deliberately chosen to provide historical perspectives and geographical balance, gender balance was apparently not considered important [13]. Other takes on OR chronology note 170 researchers of historical importance, with only 2 being women [14]. Things do seem to be on the improve however, with recent estimates suggesting approximately 26% of OR practitioners are female [15]. Obviously, like much of Science Technology Engineering and Mathematics (STEM), OR has a long way to go.
[Music plays and fades]
Mish Kardachi: What an interesting synopsis.
If you want to learn more about DST’s research and how it could be relevant to the wider science or even sporting community follow us with @DefenceScience on Twitter, or download the DST App from Google Play or the App Store.
The Defence Science and Technology podcast is a production of the Defence Science and Technology Group, part of Australia’s Department of Defence. That’s all for now. See you next time.
ENDS
References:
1. Operations Analysis: https://www.dst.defence.gov.au/research-area/operations-analysis. [Accessed 24/3/2017].
2. Tzu, S. (1910) The Art of War (translated by Lionel Giles, M. A.)
3. Ravindran, A., Phillips, D. T. and Solberg, J. J. (1976) Operations Research Principles and Practice, John Wiley and Sons
4. Trefethen, F. N. (1954) A history of Operations Research. Operations Research for Management 1 3-35
5. Falconer, N. (1976) On the Size of Convoys: An Example of the Methodology of Leading Wartime OR Scientists. Operational Research Quarterly (1970-1977) 27 (2) 315-327
6. Donovan, P. (2006) Pyrmont People: 50 Years of Defence Science in Sydney 1956-2006, Defence Science and Technology Organisation
7. Bland, L. M. (2001) A History of the First Fifty Years (1940-1990) of Support to the RAAF by Materials Scientists and Technologists in Aeronautical Research Laboratories at Fishermens Bend, Melbourne, Australia DSTO-GD-0300,
8. Donovan, P. (2007) Anticipating Tomorrow’s Defence Needs, A Century of Australian Defence Science, Defence Science & Technology Organisation
9. West, M., Cooper, T. and Kachoyan, B. (2010) AIS Analysis in Support of Counter-Piracy Operations. Australian Journal of Maritime and Ocean Affairs 2 (4)
10. Lewis, M. (2004) Moneyball: The art of winning an unfair game, W. W. Norton & Company
11. Kachoyan, B. and West, M. (2016) Cricket as life and death. In: Australasian Conference on Mathematics and Computers in Sport, Melbourne, ANZIAM MathSport
12. Rand, G. K. (2000) IFORS: the formative years. International Transactions in Operational Research 7 (2) 101
13. Rand, G. K. (2006) IFORS' Operational Research Hall of Fame. International Transactions in Operational Research 13 (6) 583-584
14. Gass, S. I. and Assad, A. A. (2005) An annotated timeline of operations research: an informal history, Kluwer Academic Publishers
15. Raynard, J. C. (2012) IFOR Survey of Global OR Practice Final Report.