Precise distributed clock synchronization is important for Time-Triggered Ethernet (TTEthernet) in which fail-arbitrary synchronization failure cannot be treated by existed protocols unless every synchronization node is equipped with a dedicated clock monitoring. A novel method was presented to detect arbitrary synchronization failure by some application-level routines which make distributed decisions mainly by monitoring the protocol control frames (PCFs). An arbitrary failure node can be exactly detected and located by the routine resided on each of the Compression Masters based on the concerned accusation messages sent from Synchronization Masters or Synchronization Clients. The proposed method can make up the weak points on the detection of the arbitrary failure node of the existed fault-tolerant synchronization mechanism and enhance the synchronization node to resist the arbitrary failure for TTEthernet. By our SAL-based model checking, this method had been formally verified to have a fail-arbitrary detection capacity even in a standard integration configuration. Simulations imply that the quality of the whole clock synchronization is effectively improved after the failure node has been isolated.