I'm dealing with a study that compares three different daily regimens of one drug in same patients (morning, noon, evening). Blood parameter that varies with the drug intake will be measured and compared between different regimens.

Repeated measured ANOVA / multiple paired T tests can be used for this purpose. However, I have a problem with determining optimal sample size for such study.

I have no idea how to calculate needed sample size for rmANOVA, but I have found different online calculators for paired T test that give me back different information (from 11 to 69).

I would like to determine a difference of 1.0 between any of the groups, standard deviation of such difference is unknown, but this parameter usually has a value between 2.0 - 4.0 and standard deviations between 2.5 and 3.0 in different samples/measurements, correlation coefficient between measurements would be 0.4 according to my pilot data. Also, power required would be at least 80%, alfa level for paired T test should be set to 0,016667 (three comparisons Bonferroni corrected 0,05).

I would be most grateful if someone could help me with this calculation or at least provide me an advice. I hope I have given enough information for these calculations.
Thank you in advance