I got the idea for this experiment from a great Science Buddies page, which suggested using Lego bird models to investigate the optimum position for flying in a V-formation.
I hypothesised that the optimum position for the leading bird (that with the least amount of drag on the following bird) would be quite close to the following bird, at about a 45° angle.
I decided to base my experiment around the Great White Pelican (Pelecanus onocrotalus), because I was able to find papers on both its body dimensions (for creating a scale model), and also the energy savings of this species flying in formation.
The model birds I used are shown below. I used a 1:20 scale, as these pelicans have a wingspan of about 2.5 metres.
I placed the following bird on a hinge (illustrated below), which was located at one end of the baseplate. The hinge rested on my digital scale (accurate to 0.01g).
I used a desktop fan to provide a wind source, but didn’t use a wind tunnel, as I thought this might simulate the stochastic environment of the air a little more. It would be very easy to put up a large box on each side, and one on the top, to make a simple wind tunnel if more control over the flow was required.
The results, shown in centigrams (0.01g) are plotted on the figure below. The position shown is the position of the leading bird, while the position of the following bird remained fixed at (8, 0). The x-axis shows the distance across from the left side of the baseplate to the leading bird (in units of 2 lego stud bricks, about 16 mm), and the y-axis shows the distance forward from the centre of the following bird to that of the leading bird.
The analysis below shows the same graph, but coloured according to the weight (which corresponds to the amount of drag).
There was obviously a lot of errors in this experiment, and there are no clear-cut results. However there are several areas where minimal (or no) drag can be seen, such as at (2,3) (8.5, 4), and (4,5). I am inclined to think that the point at (4,5) may be an anomaly, but that at (2,3) is probably a significant result, as there are neighbouring points with lower (10-20cg) weights.
Further testing is required in this area, and there are plenty of opportunities to improve the setup. There are probably a lot of simple solutions to remove most of the errors, but there will inevitably be some fluctuation in the weight.
It would also be relatively simple to measure the angles of the lever system and use the weight to calculate the force of the drag, using the formula for angular force. This would improve the usability of the results. If you decide to do this experiment, let us know how it goes!