Leidenfrost drops are liquid drops levitated by their vapour above a superheated substrate. They are well known for their spectacular mobility, which has traditionally been attributed to the lack of direct contact between the drop and the substrate. Recent experiments, however, have revealed that Leidenfrost drops can roll and move spontaneously —  in the absence of external gradients and geometrical asymmetries [1].
Aiming to illuminate these findings, we theoretically investigated the dynamics of Leidenfrost drops based on a simplified two-dimensional model, focusing on small (quasi-circular) droplets [2]. The model couples the equations of motion of the drop, which flows as a rigid wheel, and instantaneous thin-film equations governing the vapour layer below the drop.
We found that the stationary-symmetric state of a two-dimensional drop is unstable above a critical radius, leading to symmetry breaking of the liquid-vapour interface and the flows within the drop and vapour layer, and concomitant spontaneous motion of the drop. Overall, the predictions of the model are in good qualitative agreement with the experiments. Furthermore, the theory appears to rationalise several key experimental observations, such as the origins of the measured self-propulsion force and its correlation with the asymmetry of the interface.
To bring the theory closer to reality and thus allow quantitative comparison with experiments, we have recently extended our model to three dimensions, and I will discuss preliminary results based on this extension.
[1] A. Bouillant et al., Nat. Phys. 14, 1188 (2018)
[2] R. Brandão and O. Schnitzer, Phys. Rev. Fluids 5, 091601(R) (2020)

Getting here