Modelling a Better World
Professor Dylan Jones was on the wrong side of the wall.
From inside the festival grounds, he could hear music carried on the breeze – so close, and yet so far. There on the outside, sweltering in the heat, he had been queueing for two long hours.
It didn’t have to be this way. And for some people, it wasn’t. Other queues were moving much faster. He could see exactly what had gone wrong. And how to fix it.
In fact, he could have stopped the problem in its tracks. Because during the working week, that’s the sort of thing he does – and on a much bigger scale, with far higher stakes. In fact, some of his work has life-or-death implications.
Dylan is a Professor of Operational Research at the University of Portsmouth. Put simply, that’s the science of better decision making.
His main focus is multi-criteria decision analysis. This helps people within businesses and other organisations to make decisions where there are multiple criteria, and perhaps a set of goals that conflict with one another.
Dylan explains, “These ‘criteria’ are ways by which the goodness of any solution can be measured.”
It’s about making a decision in a systematic, justifiable way, that somebody from the outside could look at and see why the decision was taken the way it was. Investments rely on a level of confidence, so you need to prove things will work. It’s difficult to do that unless you test it out.
Dylan develops mathematical or computational models which are designed to address specific challenges. Every decision has its own quirks, which need to be transcribed into mathematics and built into a bespoke model. He explains:
“It’s about making a decision in a systematic, justifiable way, that somebody from the outside could look at and see why the decision was taken the way it was.
“In most decisions, there’s an element of subjectivity. But it’s important to get the person to understand what kind of weightings or preferences they’re putting on each of the criteria.”
This is not about taking decision making out of the decision maker’s hands. It’s not a case of automating hard choices. Instead:
“We guide the process by presenting a set of alternative solutions to decision makers. If they don’t like a solution, they can look at the criteria to see why. Does it cost too much? Does it not provide good customer service?
“They can then move to another decision mathematically, which may satisfy. In effect, they’re performing a trade-off between different criteria. We help them to understand the dynamics of the decision they’re making.”
We guide the process by presenting a set of alternative solutions to decision makers. If they don’t like a solution, they can look at the criteria to see why. Does it cost too much? Does it not provide good customer service?
Years of experience meant that Dylan could see how, at the music festival, criteria had not been investigated when deciding on the queueing process.
“There were parallel servers. The technology one server was using to manage entry had stopped working. Instead of diverting people to another server on a first-come-first-served basis, the queue was allowed to grow.
“Meanwhile, some people were freely jumping to shorter queues while others stuck to the rules. There was a wave effect which went backwards through the queues.
“If the event manager had looked at their decisions in a systematic manner, this could have been dealt with in around 20 minutes, rather than 2 hours.”
Dylan’s work has a lesson to teach people in all kinds of scenarios:
“It would be good if people making decisions thought through the reasons why more. A basic principle of operational research is to divide things into factors in and out of our control.
“Sometimes, the best decisions come about because you hadn’t realised exactly what could come under your control.”
Dylan’s research spans areas that are far more complex than queueing – from renewable energy and sustainability, to healthcare.
New avenues for cancer treatment
Dylan is making important contributions to a European Union funded project, which ultimately aims to increase survival rates among prostate cancer patients.
The Cooperative Brachytherapy project (CoBra) has two key strands. The first looks at the geographical placement of centres for treating prostate cancer.
Dylan and his team are mapping the UK to see how far people have to travel for treatment.
The data will show up ‘cold spots’ where people have to go a particularly long way or face especially challenging journeys.
In assessing accessibility in different areas, the mapping will look at more than just location. Other factors could include income distribution, age and other socioeconomic criteria. This ‘descriptive analysis’ will reveal whether particular groups or areas are at a disadvantage in terms of accessing treatment.
I like to work on projects that benefit people. The idea of having more efficient treatments is a real driving force.
Strictly speaking, this is the remit of the project. But Dylan, who has lost people he knew to prostate cancer, imagines that he will feel driven to go a step further and deliver his own ‘predictive analysis’ (looking at the future implications of the current situation if unchanged) and ‘prescriptive analysis’ (recommending changes and outlining the benefits).
These further analyses could potentially influence decisions for new treatment centres being built, or innovative solutions such as mobile treatment infrastructure.
The other key strand of the project concerns the introduction of robots to deliver a ground-breaking new approach to treatment.
Currently, treatment for prostate cancer involves a course of 20 or 30 injections. The new robot would deliver just one, with a payload of 100 radioactive seeds deposited inside the prostate. This would significantly reduce trauma to the patient, and speed up recovery times.
So where does Dylan fit in? As well as mapping the locations where robots would make the greatest impact in treatment centres, he is using his mathematical expertise to help plot the robot’s trajectory through the tissue, to the prostate. These are very different applications, but the maths is fundamentally the same.
By 2025-30, this research could have helped to revolutionise treatment and save many lives.
“I like to work on projects that benefit people,” says Dylan. “The idea of having more efficient treatments is a real driving force.
“It’s the same with my work on renewable energy. As an island nation, we can generate our own energy with wind and waves. I think the whole principle of science and technology is to push into the challenges and make it happen.”
A brighter future for renewable energy
Research by Dylan and colleagues at the University of Portsmouth has contributed to new ways of thinking about renewable energy.
“We worked on an EU project called LEANWIND, on offshore wind farms. Our research showed that, if things are done correctly, you can generate electricity just as cost effectively from wind farms as you can from, say, a nuclear or gas power station.
“When we started, there was huge scepticism, but now they’re saying that the big offshore wind farms in the North Sea are generating power more cheaply than nuclear.”
Speaking to many big energy companies at a large European consortium, he led work looking at how to make wind farms more cost efficient. He worked in collaboration with civil engineers and experts in subjects such as marine vessels.
The mathematical models his team developed were designed to have a positive effect on life cycle costs. This is likely to have had an impact on lowering the price of energy.
Dylan was invited to be a key investigator on LEANWIND because of his team’s eye-catching previous research into the factors – such as routing and scheduling – which would make a difference in solving challenges around wind farm maintenance.
The only way you’re going to prove it is by mathematical modelling. You’ve got to provide sufficient details of the operations, maintenance, logistics, supply chains, all of the environmental and social modelling.
He is now turning his attention to other forms of renewable energy.
“There’s real potential for tidal energies in the Channel, near France and off the coast of the Isle of Wight. So now we’re having to make the case for tidal power.
“We need to look at all the same things we looked at for wind farms – how devices work and how they’re maintained, how you get the electricity back to the shore, and so on. We’ll try to show how it’s possible to drive down the cost of tidal energy.”
Dylan is clear that advances in sustainable power rely strongly on research like this:
“There’s an element of chicken-and-egg. Investments rely on a level of confidence, so you need to prove things will work. It’s difficult to do that unless you test it out.
“The only way you’re going to prove it is by mathematical modelling. You’ve got to provide sufficient details of the operations, maintenance, logistics, supply chains, all of the environmental and social modelling.
“That can prove something will work before it’s built. It’s breaking the cycle of not being able to learn because you’re not constructing it.”
This kind of research has the potential to underpin far-reaching, positive changes – from supporting job creation, through to lowering energy bills and, of course, reducing the pollution that feeds climate change. It is genuinely life-altering.
Not perfect, but right
How does Dylan’s approach to modelling work? Let’s take tidal power as an example:
“Tides are already well-mapped and we can get a rough estimate of how much power would be generated.
“But then you might not have so much information on the future social impact, or it might be uncertain, so you’d bring in different specialist partners – geographers, doctors, engineers, for example – and bring together all their data.
“This takes us to a meta-modelling stage where we can provide criterion values, such as the various life cycle costs of building in different locations.
“We then put those in a multi-criteria decision making model, that will start to make decisions about where or when or how much you should build, what type of turbines you should use, what your supply chain route would look like, and so on.
You might not have so much information on the future social impact, or it might be uncertain, so you’d bring in different specialist partners – geographers, doctors, engineers, for example – and bring together all their data.
“You reconcile the data at a high level to try and get these criteria trade-offs and good decisions. Then you can go back to your experts who have detailed understanding of the sub-criteria, to discuss the results the model is showing.”
So, although this process is based on mathematical modelling, it is at heart a very human endeavour. As Dylan says:
“I think there’s a myth, that people think of these kind of things as some sort of artificial intelligence, where you put it all into a computer and a perfect solution pops out.
“In reality, all the good modelling processes I’ve seen in operational research have been more of an interaction with the decision maker or makers, which helps them to make a good decision based on these criteria. There is no ‘perfect’ solution because it’s a trade-off between conflicting criteria.”
It’s not a case of how to get from A to Z. It’s much more complex than that. Imagine instead that you have a choice to start at A, B, C or D.
Using visualisations such as graphs, spider diagrams or geographical information systems solutions to present the data, the model shows you what would happen if you started from those different points, and then took route E, F, G or H, prioritising factors I, J, K or L.
It doesn’t tell you what decision to make. But it helps you to make the right trade-offs on the way to reaching your own conclusion.
In reality, all the good modelling processes I’ve seen in operational research have been more of an interaction with the decision maker or makers, which helps them to make a good decision based on these criteria. There is no ‘perfect’ solution because it’s a trade-off between conflicting criteria
To the Arctic and beyond
For Dylan, the draw of operational research has always been the way it enables him to use his talent for maths and make a difference in the world.
“When I first came across operational research as an undergraduate, it appealed because it was taking real life, writing it down as a series of algebraic equations and, by solving those, producing results for somebody that will change something.”
For three decades, Dylan has been filtering real life through maths to make it better. He feels the University of Portsmouth is the ideal place to do this, because of the focus on applied research, and support and flexibility for pursuing his own agenda.
And where will his enthusiasm take him next?
“I want to look at Arctic logistics and Arctic search-and-rescue. The ice is melting up there, which means more shipping traffic can pass through. That increases the risk of accidents.
“Consider that some tourist ships carry thousands of people across thousands of miles through Arctic waters. We’re going to work with the Norwegian coastguard and the Arctic nations, to look at where you locate your facilities, how you get emergency helicopters out, and how to do all that while maintaining a fragile ecosystem.”
Once again, Dylan Jones is ready to prove that maths can be a matter of life and death.