I am not sure how something like fmriprep would use that information - most software, such as FSL, SPM and AFNI (with reverse phase encode images rather than fieldmaps) make sure of just one fieldmap to correct all of the data acquired during an acquisition. It is possible to put together your own processing pipeline that would make use of multiple fieldmaps, but I don’t think it would be that useful. I could only imagine doing something like that for very long scanning sessions (>1hr) at 7T, for example.
I tend to put a fieldmap in the middle of the scanning session, to give the participant a break, and to ‘split the difference’ of field changes and head position.