I gave an overview of the research I’m involved in in a previous post. In a nutshell its about looking at an agile design process for UAVs in the context of Search And Rescue (SAR) missions.
Part of the research involved building a detailed operational simulation model of the Solent area on the South coast of the UK. This simulation model is seeded with historical SAR incident information, weather patterns, as well as the locations of coast guard/RNLI stations & their capabilities. By running the simulation we can then analyze the spatial and temporal distribution of incidents, what the response times are, how often helicopters & lifeboats are needed, etc.
The goal is then to use those results as a baseline and see how things can be improved by adding a UAV capability to the mix. Intuitively you would think you should be able to save more lives at a lower cost by replacing (some of) the expensive helicopter/lifeboat usage by low cost UAVs.
To answer this question in any kind of reliable way requires a lot of different codes, ranging from concept design & sizing tools to high fidelity CFD and cost models. I wont go into detail but just wanted to show some of the plots produced recently.
As you tie things together you can start generating plots like this:
Which shows how the number of lives you can save over a five year period (if I remember correctly) varies in function of the landing speed and payload weight. The landing speed is a good surrogate for the overall performance of an aircraft and you can think of an aircraft with low landing speed to be a large, slow plane, while a high landing speed entails a small, fast aircraft. The payload weight represents the quality of the camera. A big, expensive, heavy camera will have a higher probability of spotting incidents in the water than a small, lightweight, cheap camera. So as you would expect you save more lives with a better camera and a faster aircraft (the latter dependency is not so strong but there are reasons for this).
Taking just one point on the above surface (so you pick a particular UAV design) you can then translate this into a survival probability and plot this on a map:
There are lots of details and assumptions behind these results (e.g., the irregular pattern comes down to the individual type of incident) and they are not final by any means. However, they are proving very useful in sanity checking our codes and pointing at areas of future work. The question we ultimately want to answer is: what are the capabilities a UAV must have to be useful in a SAR context and what is a good (optimal?) fleet mix. Still lots of work to do but promising so far.
Now just to convince my boss that we should compete in Outback Joe next year 🙂