DDDAS Approach to Fire and Agent Evacuation Modeling: Case Study of Rhode Island Nightclub Fire

1
DDDAS Approach to Fire and Agent Evacuation
Modeling Case Study of Rhode Island Nightclub
Fire

• Alok Chaturvedi
• alok_at_purdue.edu
• Jay Gore, Sergei Filatyev, Angela Mellema, Tejas
  Bhatt, Chih-Hui Hsieh
• Purdue Homeland Security Institute
• Purdue University
• West Lafayette, IN 47906

This research was funded in part by NSF DDDAS
grant CNS 0325846 and Indiana 21st century
Research and Technology Fund.
2
The Team

• Alok Chaturvedi, PI, Roko Aliprantis, Steve
  Beaudion, Jerry Busemeyer, Daniel Dolk, Jay Gore,
  Elias Houstis, Rich Linton, Shailendra Mehta,
  Suresh Mittal, and Rusi Taleyarkhan. Eric Dietz,
  Anant Grama, Chris Hoffman, Ahmed Sameh, Vernon
  Rego, Jenna Rickus, Jari Niemi, Chih-hui Hsieh,
  Teja Bhatt, Angela Mellema, Beth Naylor, Cliff
  Wojtalewicz, Rashmi Chaturvedi, Chee Foong, Midh
  Mulpuri, David Lengachar, Yeeling Tham, Steve
  Mallema, Johnson Char
• State of Indiana
• Indiana National Guard
• Purdue University, West Lafayette, IN, USA
• Naval Postgraduate School, Monterey, CA, USA
• Indiana University, Bloomington, IN, USA
• Simulex Inc.

3
Continuously Running Simulation
Automated xNA-NG update systems
JSAF
SWS
Legacy Database
Results of the simulations updates the xNA
Semantic Mining Engine
SEAS-NRT
CNN
Society of simulations
Blogs
xNA provides starting conditions and behavior to
the society of simulations
Ontology Engine xNA
The xNA is automatically being updated with
recent world incidents
Analysts and domain experts update the xNA in a
computer understandable representation
Interagency User Communities
Analysts
4
Why Integrate Simulations

• To broaden deepen the scope of simulation-based
  modeling.
• Enables diverse simulations to contribute to
  larger, more complex problemss.eg.,
  StructureSim, FireSim, HumanSim? more than
  just building engineering, physics of fires, or
  evacuations in isolation.
• Static approximations in simulations can be
  replaced with dynamic data produced by other
  simulations.eg., Heat on building structure
  comes from fire burning paper-based products in
  the building.eg., Flow of smoke is affected by
  escapees opening doors windows.

5
The Need to Address Heterogeneity

• Diversity of models leads to diversity of
  simulations.
• Two Approaches
• Standards-Based Force all simulations into a
  predefined integration architecture.
• Distributed-Development Enable interactions
  among independently developed simulations through
  translations.

6
Benefits of Distributed Development

• Simulations can be extended without interfering
  with the development of other simulations.
• Simulations can be added/removed without
  requiring updates to other simulations.
• Decoupling of producers and consumers.
  Dependences emerge at run time.
• Simulation-specific synchronization mechanisms
  and granularities.

7
How to Integrate Simulations

• metaphormembers cooperating in a society
• Simulations are autonomously managed members.
• Members contribute to societal goals by achieving
  personal objectives.
• Certain aspects of modeled reality are shared by
  multiple members.
• The same information can be interpreted
  differently by different members.

8
A Society of Simulations Shared Reality

• Accessible by all members.
• Persistent data. Members access data through
  pull-based sensing.
• Data storage can be distributed.
• Light weight data exchange.Low overhead for
  managing the members.
• Does not perform translation between consumers
  and producers.
• Does not keep a map of consumers to producers.

9
A Society of Simulations The Member

• Analyze its inputs and potential outputs with
  respect to the purpose of the society.
• Each input/output is described by its syntax,
  granularity (temporal/spatial/etc.), and
  semantics.
• Describe dependences between the member and its
  inputs.
• Use member description to build an adaptable
  bridge.

10
A Society of Simulations The Bridge

• Performs data exchange between members
  inputs/outputs and shared reality.
• Translates syntax of data in shared reality into
  a form digestible by the member.
• Interpolates/reduces across granularity
  differences.
• Handles member-specific run time issues roll
  back, checkpointing, ....

11
A Society of Simulations The Bridge

• Ontological Matching
• Knowledge discovery of information in the society
  applicable to the members inputs.
• Interprets semantics of the data according to
  members ontology.
• Extends societys ontology to enable members
  concepts.
• Links emerge as a members bridge
• Locates and obtains dynamic input data at run
  time.
• Synchronizes the member with the run time data it
  consumes.

12
Spatial Granularity of Models
Infrastructure
Economy
Markets
Political Inst
Gov Inst
Security Inst
Mil Inst
E Inst
S Inst
Info Inst
Infra Inst
Networks and Supply Chains
Social Networks
Leaders
Soldiers
Citizens
Crowd
Space
Neigh.
City
Province
Country
Region
Locale
13
Time Granularity of Models
Infrastructure
Economy
Markets
Networks and Supply Chains
Social Networks
Leaders
Soldiers
Citizens
Crowd
Time
Seconds
14
Evacuation

• This simulation was modeled after The Station
  Nightclub fire of West Warwick, RI on February
  20, 2003
• There were over 400 people present in the
  building the night of the fire
• The fire spread rapidly in just a few minutes
  the building was engulfed in flames

15
Evacuation

• Starting with the extensive research NIST has
  done on fire spread, further research was
  completed to incorporate agent behavior during
  the fire
• Agents choose evacuation routes based on their
  location and previous knowledge of the building
• Agents are affected by the smoke, CO, and CO2
  levels throughout the building

16
Evacuation
After 5 Seconds
After 30 Seconds
17
Evacuation

• Fire at times t30, t70, and t76
• Shortly after 70 seconds room
• experiences flashover

18
Evacuation

• Using agent simulation and visualization allows
  evacuation-hindering obstacles within the
  building to be clearly seen, and allow building
  designers to make changes and assess the impact
  of those changes before completing a building

19
Wider Evacuation Modeling

• This type of simulation can be integrated with a
  simulation of an entire city during a crisis
  situation
• Creation of a virtual ground zero
• Aid in training first-responders for preparedness
  and response
• Test and develop strategies in responding to
  emergency situations
• See a broad picture (city/country wide) and
  results of decisions made by first-responders
• A building such as this could be put into any
  city Mixed Reality

20
First Person View for Emergency Response

• A virtual city has been created using the Unreal
  Game Engine.
• All of the major aspects of a city have been
  including
• Transportation
• Office buildings
• Schools
• Hospitals
• Residential areas
• Commercial areas
• Parks/Recreation areas
• Emergency Responders
• Agents

21
City View

• Using the Unreal Game Engine enables a
  first-person view of the crisis situation and how
  various actions affect the scenario.
• Unreal also provides an immersive environment
  where every element of the scenario can be
  visualized fire, evacuation, quarantine
  measures, roadblocks, etc.

22
City View
23
Building View

• The user of the simulation is able to enter into
  several key buildings within the city, allowing
  them to observe the crisis workflow that develops
  within the building as the crisis progresses.

24
Modeling National Airspace
25
Air Traffic Control
26
Some other ongoing work
27
On-demand Scaling within SoS
28
Architecture for a Complex Synthetic Environment
Database
X-sim
Red/White Cell
RTI
Kinetic
Entity Based
Query Service
Pull-based stats. DB Supports Multi-COA
TACSIM
CBRNE
COA Analysis
29
Red Team
Communication Model
Intervention Models
Science-based Models
Epidemiological Model (SEIR)
Organization Model
Tools Suite
Collaboration Tools
Weather
Physiology Models
GIS Database
Socioeconomic Information
Virtual National System
Courses of Action Model

Response Asset Database
Coalition Forces
Fire Model (FDS)
Agent Based Model
Logistics/ Utilities
Scenario Database
Structure Model (LS Dyna)
Response Network
Intelligence Information
Experiment Database
Imagery Overlay
Crowd Models
Chemical Model
Open Source Intel Database
Terrain/Cultural Features
Behavior Models
Radiological Model
Live Exercise Sensors
Economic Model
Process Models (HITL)
Decision Support Model
Consequence Assessment Model
Inter-agency Incident Management Group (IIMG)
Living Lab Experiment
Command Center of the Future
Access Grid
Command Center
Virtual Ground Zero
30
(No Transcript)
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)