In manufacturing Grid workflow, multiple tasks share a common and limited resource pool. In order to solve task scheduling in multi-process with resource constraints under MG workflow, the Task-Resource Constrained model is set up to descript the assignment relation between task and resource. The framework of the task scheduling and the scheduling policies are also presented that can readjust the tasks according to the priority rules and the time parameters in the process. Then we present a heuristic scheduling algorithm that includes multiple policies. The heuristic scheduling algorithm will update the critical path of DAG (Direct Acyclic Graph) and the beginning time of post-tasks. MG Workflow engine can dynamically schedule the resources according the task requirement. An example is given to illustrate the method at last.