In the field of pig ironmaking process, operation malfunction is caused by both enlarging a blast furnace and increasing use of low quality ore. Thus, operation support systems predicting a blast furnace performance are demanded. This paper reports application of a partial unsteady simulator and a "Large-scale database-based Online Modeling (LOM)" for blast furnace operation support. First, the partial unsteady simulator constructed by using steady model as fundamental model is described. Second, a generalized LOM rebuilt by the existing LOM system is introduced. Third, a Java-based operation support system is described. The system is realized by combining the simulator of the physical modeling method and the LOM of the local modeling method. The system predicts dynamic molten pig-iron temperature in blast furnace. To predict the dynamic molten pig-iron temperature, a basic tank model of metal and slag is incorporated into the simulator. Functions representing effect of pulverized coal injection are incorporated into the unsteady model.