In a workflow management system, appropriate allocation of its activities greatly contributes to improvement
of its efficiency. We have proposed OXTHAS, a load-balancing method of scheduling the activities in
workflow management systems using Web services. The OXTHAS makes the activities be allocated to appropriate
executors based on the estimation of their processing capacity using execution histories in workflow engines. In this
paper, we propose a failure-aware re-scheduling method for the OXTHAS using process timeouts under the network
and system failures. To allocate activities appropriately, the estimated processing capacity and timeout duration
are re-calculated with considering the penalty for a failure when a process timeout occurs. We then evaluate the
effectiveness of the proposed methods through simulations.