One of the characteristics of R&D "discovery-type" projects is the "trial-and-error" aspect of "design-build-test-redesign-rebuild-retest" iterations. If this sounds familiar, you might find the Critical Chain methodology useful.
Most of the writing on the approach has discussed its use of buffers to account for variability inherent in individual tasks. Buffers are also useful as a time repository for "iteration variabilty." You don't want to promise a project based on what it could take (5 iterations), but you also don't want to promise it based on getting lucky (1-2 iterations). Using buffers to help account for the other possible 3 iterations is an "schedule-efficient" way of addressing this issue.
Saving Changes...