It has been proposed that the array might be shut down during the Jan-Feb 'Bolivian Winter' in order to save costs. Transparency for median, 25% and 75% conditions may be seen here. Note that opacity during this period may be as high as 0.3 at 225 GHz, which translates to about 6mm of PWV but that in some years it is better. The phase statistics show less variation. Although submm operation may be difficult for much of the period, operation at mm wavelengths is still quite practical. Note that Simon has updated the data. I believe the most recent writeup is ALMA Memo 512.
It has been suggested that only moving the array through a subset of configurations may save personnel costs, by eliminating one transporter crew. An investigation by Rob Reid suggests a configuration could emulate its immediate neighbors without sacrificing much sensitivity, but stretching it two configurations away would be more serious. If the moves were cut to 12-15, or skipping one configuration (and perhaps alternating on the inward cycle) the investigation suggests little loss. Larger savings could be realized by, for example, moving through fewer configurations--perhaps say eight in one direction, with moves in the other direction chosen to move through complementary arrays.