The concept behind six sigma is that you accumulate tons of data to determine the standard deviation of a process (eg dimension, material property, etc.) Its objective is admirable - back up management decisions with hard data instead of "seat of the pants" reactions. It works best when you are manufacturing thousands of widgets.
But in a field situation, the only way to accumulate this much data is to measure everything, inspect everything, log everything, over and over. It's VERY labor intensive over and above normal ops.
So I feel it could well be counter-productive to implement six sigma, if the objective is field cost reduction.