In view of the rapid large scale development of underground in the central big cities, especially around terminal stations, evacuation experiments have been conducted with 79 subjects using an underground shopping mall near Tokyo Station actually in service after business hours. Four groups of 20 subjects started from different locations and time to “escape” from any of emergency exits was measured for each subject. 12 out of the 79 subjects wore elderly simulator to compare evacuation behavior of elderly and that of younger people. The evacuation behavior was also monitored by video cameras and escape route of each subject was reproduced afterwards. Five experiments were conducted using the lighting conditions and layout of emergency exits as parameters. The starting point was changed from experiment to experiment, not making any subject repeat similar selection of escape route. The shopping mall is composed of grid of passages, and is installed with escape route sign lights on every passage according to Fire Service Law. In every experiment, it was observed that considerable portion of subjects did not select right passage toward nearest exit at intersection nor passed over right exits. These are not compliant with widely accepted assumption for deterministic modeling of evacuation in mall-like facility. Factors dictating the probability to select right passage at intersections and that to recognize emergency exit are analyzed from the experimental data.
