The effective education from an expert to a newcomer is an important issue to consider the meaning of accumulation of technical skill succession and know-how, or efficient training to a worker in the cell manufacturing system. We proposed the operator allocation method in OJT using a skill index for the cell manufacturing system of a task divide and a task exclusive system. However, in order to improve transfer and productivity of skill, the situation where two or more operators were allocated the same task occurs in an actual shop floor. In this study, we proposed the operator allocation method for collaborative operation with the procedure of task division.

In a labor-intensive cell manufacturing system, effective training is an important issue. We have proposed "skill index" as a new human index in the previous study. We have also proposed operator allocation method using this index for effective training up to now under the stable order condition. However, the order condition is fluctuating in a real site. Since this method did not change task assignment, operators had to continue the same tasks after they became expert about assigned task. It has a potential which becomes the big restrictions under fluctuating order condition. In this study, we make a new operator allocation and scheduling method for fluctuating order condition. In particular, we add a new variable for representation of quantity in decision variable of task assignment. The operators have various tasks and we have to consider about the simultaneous tasks, parallel tasks and the preceding restriction for other operators in the same cell under scheduling method. As a result of the computer experiment, our new method can get the task assignment and schedule which achieve more effective training than the previous study.

We proposed a new human index named "skill index" and also proposed operator allocation and scheduling method using this index in our previous study. We confirmed the training effectiveness of our method under stable order conditions. However, the order contents are changed during operation time in a real site. In this study, we consider the operator re-allocation procedure using our method under the order change to keep operator training and productivity. As a result of the computer experiments, we confirmed that the procedure has to be changed by a timing of order change.