In Plaxis it is possible to split a dynamic time history signal over multiple phases. This is done automatically when using the same time history signal in consecutive phases (and specifying time intervals for each of this phases which are only a part of the total duration of the signal). For example: assume we are using one signal of in total 10 sec. In phase x we specify a time interval of 5 s you might then calculate t = 0 to 5 s from the signal and in phase x+1 (also with a time interval of 5 s) you calculate from t = 5 to 10 s from the same signal.
In the above mentioned versions however it was found that with the use of a compliant base boundary an incorrect “shift” in displacements may be introduced at the start of a new phase at the base of the model. Although this shift does not seem to influence the results of the main domain, it is recommended not to split the dynamic signal over multiple phases (and thus one should use the complete signal in one phase).
Also see this article for further background information on the compliant base boundary condition.
This issue has been solved in the latest release: PLAXIS 2D 2015.00