For a given torque rating, the AC motor will almost always be physically LARGER than the equivalent torque rating in a DC design. This is particularly true for slower (< 450 rpm) shaft speeds. The reason for this is two-fold. First, the AC machine must run at a lower magnetic saturation to enable equivalent performance in terms of speed control (generally the windings are higher inductance than the DC version, since more turns and less amps). Second, the AC machine is often relying on an internal fan to circulate sufficient cooling air to maintain a reasonable temperature rise, whereas the DC machine almost always uses an externally-driven fan. What this boils down to is that the power density of the DC design is higher at lower speeds, because there's more cooling available.
For a given power rating, the AC drive will almost always be LARGER than the equivalent DC rating. This is for two reasons: the greater clearances necessary when operating at higher voltages, and the simple fact that the AC drive is really two DC drives connected back-to-back with a DC bus between them.
Yes, the DC machine can require commutator and brush maintenance on a relatively frequent basis. However, the drives can experience failures as well, which are far more costly on a per-failure basis in terms of time and parts. When looked at over the long haul - say, 20 year motor life or so - it turns out that the "maintenance" cost for each type of system is roughly the same within about 5 percent.
When brushes are chosen correctly for the environment and application, it is not unusual to see 12-24 months life from a single set. If the practice is to change some of the brushes as part of a regular preventive maintenance cycle, the actual labor hours associated with the replacement is negligible. Commutator maintenance can be a sore spot - but it is often the direct result of something else going wrong in the system (i.e. ingress of contaminants or incorrect loading of the machine).
The real drawback to "brushed" DC is speed. This is a result of the limitations on commutation, which are based on the speed at which the brush passes over a given segment of the commutator to apply current to a portion of the winding. In effect, the rotational speed provides a circumferential speed - as the commutator becomes larger in diameter, it has to run slower to keep the peripheral speed (the speed seen by the actual brush-bar interface) to within reasonable limits. Secondary limits are imposed by the mechanical tolerances for a "good" commutator; going too fast will distort the bar profile, which will ultimately lead to more rapid brush wear and film degradation.
For "brushless" DC, this is not really an issue - with no mechanical contact for a brush, the speed limit becomes one of simply holding the rotor together.
Using a VFD on an AC machine only makes sense when the process requires fairly significant speed / load changes as part of the regular operating profile. It does NOT save money when used for a steady-state load condition due to the increased losses associated with the drive itself and additional cooling fans.
The cost of a four-quadrant drive is not cheap, no matter which system (AC or DC) is used, However, a DC one is often cheaper and smaller for the same reasons as given in one of my earlier paragraphs. And four-quadrant operation is required for either reversing duty OR the ability to actively "apply the brakes" to the process.