2001MNRAS.326..126S

We present a model for steady-state winds of systems with super-Eddington luminosities. These radiatively driven winds are expected to be optically thick and clumpy as they arise from an instability-driven porous atmosphere. The model is then applied to derive the mass loss observed in bright classical novae. The main results are as follows.

(i) A general relation between the mass-loss rate and the total luminosity in super-Eddington systems.

(ii) A natural explanation of the long-duration super-Eddington outflows that are clearly observed in at least two cases (Novae LMC 1988 #1 and FH Serpentis).

(iii) A qualitative agreement between the prediction and observations of the mass loss and temperature evolution.

(iv) An agreement between the predicted average integrated mass loss of novae as a function of white dwarf mass and its observations.

(v) A natural explanation for the `transition phase' of novae.

(vi) Agreement with η Carinae, which was used to double check the theory: the prediction for the mass shed in the star's great eruption agrees with observations to within the measurement error.