The cosmological remnants of a first-order phase transition generally depend on the perturbations that the walls of expanding bubbles originate in the plasma. Several of the formation mechanisms occur when bubbles collide and lose their spherical symmetry. However, spherical bubbles are often considered in the literature, in particular for the calculation of gravitational waves. We study the steady state motion of bubble walls for different bubble symmetries. Using the bag equation of state, we discuss the propagation of phase transition fronts as detonations and subsonic or supersonic deflagrations. We consider the cases of spherical, cylindrical and planar walls, and compare the energy transferred to bulk motions of the relativistic fluid. We find that the different wall geometries give similar perturbations of the plasma. For the case of planar walls, we obtain analytical expressions for the kinetic energy in the bulk motions. As an application, we discuss the generation of gravitational waves.Comment: 24 pages, 8 figures. Minor corrections and clarification

We investigate the production of a stochastic background of gravitational waves in the electroweak phase transition. We consider extensions of the Standard Model which can give very strongly first-order phase transitions, such that the transition fronts either propagate as detonations or run away. To compute the bubble wall velocity, we estimate the friction with the plasma and take into account the hydrodynamics. We track the development of the phase transition up to the percolation time, and we calculate the gravitational wave spectrum generated by bubble collisions, magnetohydrodynamic turbulence, and sound waves. For the kinds of models we consider, we find parameter regions for which the gravitational waves are potentially observable at the planned space-based interferometer eLISA. In such cases, the signal from sound waves is generally dominant, while that from bubble collisions is the least significant of them. Since the sound waves and turbulence mechanisms are diminished for runaway walls, the models with the best prospects of detection at eLISA are those which do not have such solutions. In particular, we find that heavy extra bosons provide stronger gravitational wave signals than tree-level terms. *