Convergence of the hydrodynamical structure of the LMC 400-3 model. Left plot: maximum relative difference between hydrodynamical variables in subsequent iteration steps as a function of total number of hydrodynamical iterations. We note that kinetic equilibrium equations are iterated for each step of hydrodynamical iterations. Blue vertical strips denote a new step of a global iteration (numbered in the bottom of the plot) when the inner boundary velocity (and the mass-loss rate) is changed. Labels a–d in the bottom of the plot indicate (see also Krtička & Kubát 2017, Sect. 2): Panel a:calculation of the initial model for subcritical velocities. In the first few steps the CMF radiative transfer is not solved. Afterwards, the CMF calculations are switched on (indicated by a blue arrow). Panel b: iteration of wind model with mass-loss rate lower than or equal to the final one. Each model denoted as b has a higher mass-loss rate than the previous one. Panel c: iteration of wind model with mass-loss rate lower than the final one and inclusion of model for supercritical velocities. Here the blue arrow denotes inclusion of model for supercritical velocities. Panel d: iteration of wind model with mass-loss rate higher than the final one. After this step the model mass-loss rate decreases. The plot shows that the convergence of hydrodynamical structure in each step is stable and relatively fast. Right plot: mass-loss rate in individual global iteration steps (numbers on horizontal axis are the same as in blue labels in the left plot). The models converge to a final mass-loss rate given in Table 1.