Post by GJUK on Apr 27, 2017 22:20:28 GMT -8
NASA's Advance Supercomputing Division used a 3D scan of a modified DJI Phanom 3 and set their 4.09 Petaflop Pleiades supercomputer to model the flow of air around the craft. Even with this remarkable amount of power (4,900,000,000,000,000 floating point calculations per second - that's just under 5 quadrillion!), the simulation still took five full days to complete, generating this hypnotic rendering:
This helps demonstrate one problem with quadcopters. As a rotor slices over the four-armed body of the drone, it sends ripples of pressure across the surface (as you see with the flashes of blue and red). This limits the amount of thrust available, and can lead to a wobble that has to be counteracted by Automatic Flight Control software, the computations of which combined with micro course corrections reduce all-too-precious battery life.
The visualization also exposes why most quads create more than their fair share of noise. It is hoped that visualisations like these might help lead to new ways to redesign the audio signature of drones, while simultaneously improving stability.
Blue = Low Pressure Red = High Pressure
This helps demonstrate one problem with quadcopters. As a rotor slices over the four-armed body of the drone, it sends ripples of pressure across the surface (as you see with the flashes of blue and red). This limits the amount of thrust available, and can lead to a wobble that has to be counteracted by Automatic Flight Control software, the computations of which combined with micro course corrections reduce all-too-precious battery life.
The visualization also exposes why most quads create more than their fair share of noise. It is hoped that visualisations like these might help lead to new ways to redesign the audio signature of drones, while simultaneously improving stability.