In complex nuclear energy redundancy systems, there are many failure events that do not follow specific time distribution. For these atypical time-distribution events, traditional dynamic fault tree (DFT) methods cannot be applied directly, which has posed great challenges to reliability modeling and evaluating. In this contribution, we summarize atypical time-distribution events in nuclear energy redundancy systems and propose new modeling and evaluating methods based on DFT. To demonstrate the reasonability of the proposed methods, two case studies about make-up water pumps and emergency diesel generators are analyzed in comparison with traditional DFT. The results indicate that the proposed methods can effectively model and analyze the reliability of redundant systems with atypical time-distribution events. The proposed methods can provide useful information for optimization design of nuclear energy redundancy systems and has potential to improve the economy of nuclear power plants by relaxing overestimated unreliability.