A Review of Building Evacuation Models - NIST Virtual Library

adjusts the FED susceptibility of each occupant or occupant group, which affects whether the
person become incapacitated or can escape the building space without problem.
Import CAD drawings: Yes, CAD drawings can be imported into the model via another BRE
program, Josephine. Or, the floor plan can be drawn using a graphical user interface (GUI)
within GridFlow. The user specifies links on the floor plan that lead to the outside or another
space in the building. The user is prompted to input the link width and maximum flow
(persons/second) through the link.
Visualization capabilities: 2-D and 3-D capabilities (with Josephine).
Validation studies: The model developer states that GridFlow has undergone many runs of
simple buildings and multi-enclosure spaces for the purpose of four aspects of validation:
Component testing (routine checking of major software), functional validation (checking model
capabilities and that these are compatible with intentions), qualitative testing (comparing
predicted human behavior with expectations), and quantitative verification (comparison of model
predictions with experimental data). The developers have performed component testing and
quantitative verification, which involved simulations from simple and complex building
compared against empirical data from the SFPE Handbook 8 and other sources. Functional
validation has also been performed and limitations of the model have been identified (but not
included in the Purser report). Also, human behavior has been validated by using actual premovement
data to simulate a scenario and by comparing the model’s evacuation behavior and
time to the observed evacuation and Handbook data.
Purser discusses simulations used to examine the effects of delay time, travel time, and exit flow
capacity for various occupancies and layouts. He outlines the results of a hypothetical building
with 3 different numbers of occupants. In this work, Purser could understand graphically
whether the evacuation was driven by pre-movement time, travel distributions, or exit flow
capacity, depending upon the number of occupants in the building.
Lastly, a GridFlow simulation was described that was similar to an actual evacuation incident,
the “Sprucefield” evacuation. This included 190 occupants evacuating from a food hall.
GridFlow modeled that 99 % of the occupants would evacuate in 130 s with their similar case,
when the actual time was 140 s. Purser notes that GridFlow provided reasonable results and they
plan to perform direct simulations on the Sprucefield case, among others.
Special features:
Counterflow – Yes.
Manual exit block/obstacles – Yes, because the user can specify the destination or exit choices
for each individual or occupant group, certain exits can be “hidden” (or not given as a choice)
from an occupant group as if it does not exist.
Fire conditions affect behavior Fire conditions are implicitly incorporated. The user imports a
spreadsheet (created by the user) with speed factors and FED doses with time for each building
space.
A-32