It is of fundamental importance to investigate the evacuation process from a room with obstacles. The typical case is the evacuation of students from a classroom. Based on evacuation experiments from a classroom, the essential features of evacuee are concluded. In the original floor field model, the dynamic floor field is introduced in order to reflect the interaction among pedestrians. A pedestrian may follow the virtual trace of another one in front. The static floor field does not consider the influence of pedestrians. In this paper, the original dynamic floor field is ignored. These desks and chairs are treated as impassable and passable static obstacles, respectively. The static and passible obstacles, such as chairs, lead to the delay of movement of pedestrians. Furthermore, pedestrians are regarded as movable obstacles. The effect of static obstacles on floor field does not change with time. However, the effect of movable obstacles on floor field is dynamic. Therefore, the whole floor field is updated dynamically according to the movement of crowd. Pedestrians may try to find another uncongested path or exit when they find the crowd in front. It provides a better description of the influence of downstream congestions on upstream crowd. The cellular automaton model based on the dynamic floor field is used to investigate the evacuation process in the case of four layouts and three exit widths. The spatial distributions of evacuation time in different conditions and also the average and maximum evacuation times are obtained. Numerical simulations reproduce the evacuation process observed in the experiment quite well. The evacuation time depends on arrangement of these desks and the exit width. For a given layout, the smaller exit leads to longer evacuation time. It is found that the evacuation time does not decrease monotonically with increasing the number of aisles, which depends on the width of aisle as well. When the aisle is not wide enough, the conflict of pedestrians from both sides reduces the efficiency of evacuation. It is helpful for coping with crowd evacuation with an aisle close to the exit side of the wall. The reasons of the differences between experimental and simulation results are also discussed in more detail.