In recent years a significant body of research has been dedicated to the analysis and verification of workflows without taking into account the temporal dimension of activities involved. Once the temporal aspects of the activities in a process description are taken into account, various types of synchronization anomalies and other undesirable behavior may be detected. In this paper, we use a timed Petri nets to represent workflows and introduce a set of linear inference algorithms for the quantitative analysis of temporal constraints of workflows. We apply our models to the analysis of several workflow patterns.