2002ApJ...568..312S

Theoretical models of the remnants of massive stars in a very hot, post-red-supergiant phase display no obvious instability if standard assumptions are made. However, the brightest observed classical luminous blue variables (LBVs) may well belong to such a phase. A simple time-dependent theory of moving stellar envelopes is developed in order to treat deep hydrodynamical disturbances caused by surface mass loss and to test the moving envelopes for dynamical instability. In the case of steady state outflow, the theory reduces to the equivalent of the Castor, Abbott, & Klein formulation for optically thick winds at distances well above the sonic point. The time-dependent version indicates that the brightest and hottest LBVs are both dynamically and radiatively unstable, as a result of the substantial lowering of the generalized Eddington luminosity limit by the mass-loss acceleration. It is suggested that dynamical instability, by triggering secular cycles of mass loss, is primarily what differentiates LBVs from the purely radiatively unstable Wolf-Rayet stars. Furthermore, when accurate main-sequence mass-loss rates are used to calculate the evolutionary tracks, the predicted surface hydrogen and nitrogen abundances of the blue remnants agree much better with observations of the brightest LBVs than before.