The purpose of good database logical design is to eliminate data redundancy and insertion, deletion and update anomalies. For temporal databases, temporal schemes may be normalized by using constraints of temporal functional dependencies (TFDs) with multiple temporal granularities. However, the adoption of temporal dimension and usage of multiple temporal granularities make it very complicated to design a temporal database. Generally, the set of temporal types that can be processed by a system and involved in lots of applications, meet totally ordered relation, and the set of TFDs with a totally ordered set of temporal types is closely related to the Armstrong axioms of traditional functional dependencies (FDs). By analyzing the existing relationships between TFDs and FDs and utilizing corresponding algorithms for traditional set of FDs, some important algorithms such as membership and finite closure of attributes algorithms are proposed for given set of TFDs. These algorithms are the basis of further normalization for temporal databases.