Using maths to improve decision making

Friday, June 17, 2011

ThinkTank Maths of Edinburgh, UK, has an interesting business model for the oil and gas industry – helping people make better decisions using mathematics.

In the Macondo disaster, lives were lost because people did not understand the information they were being provided quickly enough. If there was a computer tool which would sift through all the information and prioritise the most information to serve to the crew, it is possible that lives could have been saved.

To look at it another way, there is an example of a company which uses advanced maths to help  work out which information is most helpful. Most of us use it every day, even though we probably have never thought of it in this way – Google.

There are many aspects of the oil and gas industry where people have to make decisions based on complex data. For example, operating drilling equipment, automating old oilfields, integrated operations, operating subsea vehicles (ROVs), adjusting production strategy, condition based equipment monitoring. Even the decisions about which technologies to use.

ThinkTank Maths specialises in solving all kinds of problems using advanced and creative mathematics, or where the existing mathematics can be improved upon.

 It aims to employ only people from the top 2 per cent of university mathematics classes.  As well as the oil and gas industry, it has worked in defence, banking and space.

“The idea is that the machine takes some of the burden of thinking about the right option,” says Hannu Rajaniemi, director of ThinkTank Maths.

New approaches are needed to enable efficient human-machine collaboration.

One problem with many automation systems is that there is too little visibility into what decisions are actually being made by the computer – they are a “black box”.

An example, he said, was an unmanned aerial vehicle (UAV), or “drone”, which suddenly lost height flying above the sea, creating a lot of concern for the operators.

It was only after the log files were analysed, that operators realised that it was reducing altitude to take a photograph of an object floating on the sea, (a lobster buoy in this case), as it had been programmed to do.

The ThinkTank Maths approach is to find a way to involve the user in the decision making system – the system explains to the user what action it proposes to take and why.

Another common problem with automation systems is when there are multiple alerts going off at once if something goes wrong. Think Tank Maths can build algorithms which aim to understand what is happening and just present the user with clear advice – for example, evacuation the platform now.