IMO 9: Exit flow: crowd dissipation from a large public room

1. Definition

A public room of 30 m by 20 m has four 1 m doors, two in each long wall at 4 m to 5 m and 25 m to 26 m from the corner (Fig. 3, p. 12). 1,000 persons with instant response and Table 3.4 speeds are distributed uniformly in the room, inset 2 m from the walls, and leave through the nearest door. Step 1: the time the last person leaves is recorded. Step 2: doors 1 and 2 are closed and step 1 is repeated.

Source: IMO MSC.1/Circ.1533, Annex 3, Appendix 2, Test 9, para. 15, pp. 11-12, Fig. 3. This is a paraphrase, not the text of the standard.

Reference: International Maritime Organization (2016). Revised guidelines on evacuation analysis for new and existing passenger ships. MSC.1/Circ.1533, 6 June 2016. https://edocs.imo.org/Final%20Documents/English/MSC.1-CIRC.1533%20(E).docx (Page numbers are the printed page numbers of Annex 3 (“Annex 3, page n”). Tests are cited by test number and paragraph of Appendix 2.)

2. Scenario

Scenario picture not available.

3. What to expect

Closing doors 1 and 2 roughly doubles the time to empty the room (an “approximate doubling”). IMO gives no band around two.

Our criterion: Halving the exits increases evacuation time

4. Our result

Status: PASS

Test cases of this row

• test_time_ratio: bound: agents_remaining = 0 in both runs; comparison: measured[1] (2 exits) > measured[0] (4 exits)

5. Our implementation

Deviations from the standard:

• The population is 200 agents in the area x = 0 m to 25 m, not 1,000 agents inset 2 m on all sides.
• The four exits are 1 m by 1 m absorbing squares against the right-hand 20 m wall (y = 0 m to 1 m, 5 m to 6 m, 14 m to 15 m, 19 m to 20 m), not doors in the two 30 m walls as in Fig. 3.
• Agents do not choose the nearest door. Horizontal bands of the start area are assigned to one exit each through journey weights: four bands of 50 agents with four exits, two bands of 100 agents with two exits.
• With two exits, the test keeps the exits at y = 0 m to 1 m and 19 m to 20 m on the same wall. In the standard the remaining doors 3 and 4 lie in one wall at 4 m to 5 m and 25 m to 26 m.
• The pass rule is the suite’s own: both runs complete and the two-exit time is longer than the four-exit time. The ratio to the standard’s doubling is not tested.
• Constant desired speed 1.2 m/s instead of Table 3.4 speeds; default body radius 0.15 m; one seed (42). The scenario is built in code (tests/vv/test_imo.py).

Model: CollisionFreeSpeedModel

• Test function: tests/vv/test_imo.py::test_time_ratio

6. Our notebook

No notebook for this test yet.