Business Process Management Systems (BPMS) is a complex information system that provides designing, administrating, and improving the business processes. Task allocation to human resources is one of the most important issues which should be managed more efficiently in BPMS. Task allocation algorithms are defined in order to meet the various policies of organizations. The most important of these policies could be reducing the average cycle time, balancing the resource workload, increasing the product quality and minimizing the production costs. Therefore, choosing an appropriate resource in task allocation algorithms could influence on overall policy of the organization. In heavy load conditions or when the number of human resources is limited, workload balancing can increase the stability of the system. In this paper, a task allocation algorithm is proposed to rebalance the resource workload in runtime while minimizing cycle time by offering dependent pair tasks to resources for concurrent processing. The experimental results show that the combination of previous algorithms with the proposed algorithms would have 4.42% reduction in cycle time in contrast to most efficient state-of-the-art algorithms.