Coupling between human behaviour and combustion: multi – agent and CFD simulations for road tunnel fires.
Msc- Thesis by Giordana Gai
School of Engineering, University of Rome La Sapienza
Coupling human behaviour and combustion simulations for tunnel fire safety
1. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Candidata:
Giordana Gai
Relatore:
Prof. Ing. Franco Bontempi
Correlatore:
Ing. Filippo Gentili
Facoltàdi Ingegneria Civile e Industriale
Corso di laurea magistrale in Ingegneria Civile
2. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
STRUCTURAL
SYSTEM
SAFETY
NON – STRUCTURAL COMPONENT
STRUCTURAL COMPONENT
PEOPLE
1 / 18
INTRODUCTION
3. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
NON – STRUCTURAL COMPONENT
Tunnel
INTRODUCTION
Human behaviour
2 / 18
STRUCTURAL
SYSTEM
SAFETY
STRUCTURAL COMPONENT
PEOPLE
ACTION : Fire
4. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
INTRODUCTION
Mont BlancTunnel fire(1999): 39 people died
(11.6 km long single –bore bi –directionaltunnel)
3 / 18
Tunnel system complexity
Fire
Trafficconditions
Geometry
Safetymeasures
Occupant
-1 dimensionlongerthanthe othertwo
-Confinedand closedenvironment
-Cross sectiontype(n° oftubes)
-Longitudinalslope
-Dailytrafficvolume
-% HGV
-Detection system
-Alarmsystem
-Sprinklers
-SOS stations
-Emergencyexits
-Lay–by
-Illuminationsystem
-Vertical signage
-Horizontalsignage
-Numberofusers
-Composition(gender and age)
-Psycologicalcharacteristics
-Habitualusers(familiarity)
-Humanbehaviour
-Disabledpeople
-Fireload
-HRR
-Numberofvehiclesinvolvedin the accident
-Typeoffuel
-Position ofthe accident
Multidisciplinaryand multiphisicsproblem
5. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
Spatialaspects
1D -aspects
Temporalaspects
Codeand computationalcost
RANS
ReynoldsAverageNavierStokes
ANSYS FLUENT
20 cm –mesh
8 hoursofsimulation
LES
LargeEddy Simulation
FDS
20 cm –mesh
1hourofsimulation
0
1
2
3
4
5
6
7
8
9
10
0
200
400
Y [m]
Temperature [°C]
0
1
2
3
4
5
6
7
8
9
10
0
200
400
Y [m]
Temperature [°C]
0
50
100
150
200
250
300
0
5
10
15
20
Temperature [°C]
Time [s]
0
50
100
150
200
250
300
0
5
10
15
20
Temperature [°C]
Time [s]
+ Evac
COMPUTATIONAL FLUID DYNAMICS
4 / 18
6. ClassicalApproach
LPHCActions
Low ProbabilityHigh Consequences
Normative
•Structuraland unstructuralpermanentload
•Antropicload
•Snow, wind, seismicaction
Combination(ULS) G1G1+G2G2+PP+Q1Qk1+
+Q202Qk2+…
Combination(SLS)
G1+G2+P+11Qk1+22Qk2+…
A –Firecharacteristicsandsafetymeasures
•Positionofthe accident
•Fireload(HRR)
•Ventilationconditions
•Presenceand position ofsafetymeasures(by–passes, signage, illumination, sprinklers, detection systems, etc)
B –Trafficcharacteristics
•Trafficflow conditions
•PresenceofHGV
C –People
•Numberofpeople
•People composition(physicsand psycologicalcharacteristics)
•Humanbehaviour
•Panicand familiaritywiththe structure
•Presenceofdisabledpeople
ANAS Circular2009 (basedon D.Lgs264/2006 and Directive2004/54/EC)
Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
SCENARIO –BASED APPROACH
5 / 18
Fixeddata
Variablesdata
Scenarioanalysisisaprocessofanalyzingpossiblefutureeventsbyconsideringalternativepossibleoutcomes,includingthedevelopmentpathsleadingtothem.
7. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
HUMAN BEHAVIOURAL MODEL
6 / 18
PEOPLE INSIDE THE TUNNEL
Situation
Ideal (theoretical)
Real
Tunnel environmentand itsinfrastructures
Complete knowledge
Partialknowledge
Psycologicalfeelings
No fear
Panic
Behaviour
Perfectlyrational
Imperfectlyrational
Forexample, failurein usingthe safetyequipment
Forexample, uncorrectexitdoorselection
FDS + Evac
Partiallybehaviouralmodel
•Movementalgorithm
•Social force
•Panicmodel
•Reducedvisibilityforsmokeand obstacles
•Reducedvelocityforsmoke, incapacitationand death
•FractionalEffectiveDose concept
•Tendencytoactindependentlyor toformgroup
•5 categoriesofhumanswithdifferentphysicalcharacteristics
•3 behaviouraltype(rational, conservative, herding)
8. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
Sicily, Catania –Syracuse Motorway E45
THE ST. DEMETRIO TUNNEL
7 / 18
Syracuse–Cataniatube,L=2948m
Infrastructures and devices respecting the European standards
for tunnel longer than 3 km (Directive 2004/54/EC)
Catania
Cortesyof Eng. Luigi Carrarini, ANAS
Catania–Syracusetube,L=2895m
705 m
9. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
40 m
165 m
175 m
245 m
80 m
MESH 1
25x25x25 cm
MESH 2
50x50x50 cm
MESH 2
50x50x50 cm
MESH 3
100x100x100 cm
Jet fan
Jet fan
Jet fan
Jet fan
Jet fan
Emergency exit
Portal
Fire source
TRAFFIC FLOW DIRECTION
MESH 3
100x100x100 cm
Catania –Syracuse tube
THE ST. DEMETRIO TUNNEL: FDS+EvacMODEL
8 / 18
Emergency exit
Portal
140 m
300 m
Bus
Vehicle (1 passenger)
Vehicle (2 passengers)
Vehicle (3 passengers)
Vehicle (4 passengers)
Fire source
705 m
10. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
12%
63%
25%
Agents composition
Rational
Conservative
Herding
N°tot = 100 AGENTS
•Rational : Adult
•Conservative : Male, Female
•Herding : Child, Elderly
each category has a different exit selection preference order
THE ST. DEMETRIO TUNNEL: EVACMODEL
9 / 18
FAMILIARITY CONCEPT
Knowndoorprobability
1
Portal
0.5
Emergencyexit
Emergency exit
Portal
140 m
300 m
Fire source
TRAFFIC FLOW DIRECTION
11. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
ASET
AvailableSafetyEgressTime
Fromignitiontounsustainableconditionsforthe egressofhumans:
RequiredSafetyEgressTime
Fromignitionuntil the time the last occupant reaches a safe place.
It depends on: detection time, alarmtime, pre–movementtime, traveltime.
THE PERFORMANCE –BASED APPROACH FOR EGRESS
•Visibility< 10 m
•T > 60°C
•FED > 1
Pre–evacuationtime
Detection time
Reactiontime
Alarmtime
Recognitiontime
Responsetime
Decision–makingprocess
Movementtime
RSET
>>
10 / 18
12. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
0
20
40
CAR 2 LL - CAR 1 RL
CAR 4 LL - CAR 3 RL
CAR 6 LL - CAR 5 RL
CAR 8 LL - BUS RL
CAR 10 LL - CAR 7 RL
CAR 13 LL - CAR 9 RL
CAR 15 LL - CAR 11 RL
CAR 16 LL - CAR 12 RL
CAR 18 LL - CAR 14 RL
CAR 22 LL - CAR 19 RL
CAR 20 RL
CAR 21 RL
Time [s]
Formationofthe queueinside the tunnel
Left lane (LL)
Right lane (RL)
THE ST. DEMETRIO TUNNEL: EVACMODEL
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
250
275
300
325
350
Probability distribution
Time [s]
Pre -evacuation time distribution
Normal distribution
Truncated normaldistribution
ANAS ReferenceTime
ProbabilisticApproach
DeterministicApproach
11 / 18
13. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
Fire source
HRR = 6, 30 MW
Jet fans
Volume flow = 15 m3/s
Flow velocity = 15 m/sEMERGENCY EXIT
PORTAL
THE ST. DEMETRIO TUNNEL: RESULTING MODEL
12 / 18
HUMAN CATEGORY
COLOR
Adult
Blue
Male
Green
Female
Red
Child
Yellow
Elderly
Black
140
300
150
14. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
0
5
10
15
20
25
30
35
0
1000
2000
3000
Time [s]
HRR [MW]
SCENARIOS
13 / 18
Jet fans
1 Jet fan
Emergencyexit
Emergencyexit
BUS FIRE
(30 MW)
2 –CARS FIRE
(6 MW)
•Idealbehaviour
•Realbehaviour
0
5
10
15
20
25
30
35
0
1000
2000
3000
Time [s]
HRR [MW]
Emergencyexit
Emergencyexit
•Realbehaviour
15. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
RESULTS (6 MW –fire)
14 / 18
0
20
40
60
80
100
0
500
1000
N°agents
Time [s]
Ideal egress
Emergency exit
Portal
Total evacuated agents
0
20
40
60
80
100
0
500
1000
N°agents
Time [s]
Real Egress
Emergency exit
Portal
Total evacuated agentsRATIONAL BEHAVIOUR
PANIC
Ideal egress
Real egress
0
20
40
60
80
100
Emergency exit
Portal
90
10
51
49
Number of agents
Ideal egress
Real egress
0
20
40
60
80
100
0
200
400
600
800
1000
Number of evacuated agents
Time [s]
Ideal
Real
16. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
RESULTS (6 MW –fire)
0
10
20
30
40
50
60
70
80
90
100
0
200
400
600
800
1000
1200
Number of evacuated agents
Time [s]
15 / 18
0
200
400
600
800
1000
1200
1,011.0
939.0
1,120.0
1,125.0
RSET[s]
Yes Jet Fan - Yes Em. Exit
No Jet Fan - Yes Em. Exit
Yes Jet Fan - No Em. Exit
No Jet Fan - No Em. Exit
Jet Fans
1 Jet Fan
EmergencyExit
EmergencyExit
2 –CARS FIRE
(6 MW)
Jet Fans
Em. Exit
1 Jet Fan
Em. Exit
Jet Fans
Em. Exit
1 Jet Fan
Em. Exit
EmergencyExit
EmergencyExit
17. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
Human _ speed [m/s]*10-4
STREAKS
t = 593 s
t = 390 s
Portal
Emergency exit
RESULTS (6 MW –fire)
16 / 18
Thisisnotageneralresult: FDS+EvacshouldberunusingMonteCarlomethod(repeatedsamplingtodeterminethepropertiesofthephenomenon).
ThisisonlyonesamplingoftheMonteCarlosimulation.
18. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
RESULTS (6 MW –fire)
17 / 18
•Visibility _ soot [m] at z = 1.6 m
high visibility upstream the fire source during the egress time
ASET >> RSET considered scenario
•Humans FED 10-5
very low value, even for those scenarios with 1 jet fan off (emergency ventilation is so strong that there are no problems of intoxication)
•Temperature slice [°C]
low value (< 60°C) next to the emergency exit
19. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
CONCLUSIONS
18 / 18
•TheFDS+Evaccodeisagoodtoolfortheassessmentoftheemergencyegressforroadtunnelfires:fireandsmokeinfluencehumanmotionandtheireffectsimplydifferentchoicesoftheescaperoute,intoxication(untilthedeath)andreductionofthewalkingvelocity.
•ThepossibilityofusingnumericalsimulationsforegressscenariosisfundamentalfortheSafetyEngineering,becauseitispossibletoverifytheefficiencyoftheescaperoutesofmanydifferenttypesofbuilding(civilbuilding,stadium,tunnel,etc)bycontrollingthemaximumcontactforcedevelopedbetweenhumansandthemostproblematicstreaksnexttotheexits.
•ForwhatconcernstheS.Demetriotunnel,theresultsoftheanalysisaresatisfactory:forthescenariosexaminedtheemergencyegressiscompletein15–20minutes,withoutproblemsofvisibilityorintoxication.Thebacklayeringphenomenonoccursonlyforthebusfirewith1jetfanoff,butithappenswhenhumanshavealreadygoneout.
20. THE END
V –velocity field
0
20
40
60
80
100
0
200
400
600
800
1000
1200
Number of evacuated agents
Time [s]
Egress_ 6 MW fire
Ideal and Rational
Yes Jet Fan - Yes Em. Exit
No Jet Fan - Yes Em. Exit
Yes Jen Fan - No Em. Exit
No Jet Fan - No Em. Exit
12%
63%
25%
Agents composition
Rational
Conservative
Herding
Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel fires
Giordana Gai
21. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel firesGiordana Gai
BACKLAYERING PHENOMENON
Bus fire–1 jet fan off
t = 418 s
t = 594 s
t = 720 s
22. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel firesGiordana Gai
Human _ force total [N/m]
t = 300 s
t = 360 s
t = 327 s
23. Coupling between human behaviourand combustion: multi –agent and CFD simulations for road tunnel firesGiordana Gai
RESULTS (6 MW –fire)
Human _ force total [N/m]
Block in the bus