While increasingly intense global competition is becoming seriously, the management of product quality, the improvement of productivity and the cost reduction are very important issues for a company. In the iron and steel industry, they demand to realize steady operation in the continuous casting process, which is one of important process of the steel manufacturing. The decision making to choose proper operation is important by understanding a current state of the steel and the equipment in the continuous casting process and predicting the condition in the future. However, the knowledge that is necessary for this decision-making is implicit knowledge of the skillful operating worker. Such operating knowledge will disappear along with the retirement of the skillful operating worker. Problems such as increase in the disaster by the dangerous behavior and decline in the productivity have been becoming serious. A skillful operating worker has a cognitive model (a continuous casting process model) of the casting process in the inside of the head. This research tried to define and describe their cognitive model with Petri-Net, and develops an operating decision making support system for the trouble avoidance using a reachable tree in this research. Using this system, we input the steel and equipment conditions of the accident occurrence that was reported in the accident and trouble report, and predict a trouble occurrence, and extract the operation to avoid a trouble.