Structural Fire Safety State of the Art

Future Directions

Andy BuchananUniversity of Canterbury, New Zealand

Where have we come from?

1666 – Great Fire of London

1900s -• Prescriptive codes • Fire Resistance Ratings• Simple test methods• No science

Last 20 years - lots of change

What are the changes?

OUTPUTS:• Predictable

behaviour

• Better science

• Safer buildingsfor less cost

INPUTS:• Fire science

• Structural analysis

• Performance-based codes

StructuralFire

Safety

What are our objectives?

Who are the stakeholders?• Building owner• Designer• Regulator• Forensic investigator• Researchers• Manufacturers• Code writers

One building at a time

Groups of buildings

How much time have we got?

For one building:• During an emergency • Preliminary design• Full design• Risk assessment• Forensic study

minuteshoursdaysmonthsyears

What are our objectives?Life safety vs property protection?

Analysis

Design

Does better analysis give better design?

What are our objectives?

Researcher

One off designer

Everyday designer

Code writerAnalysis

What do the regulators want?

Prescriptive codes• How to build it - don’t ask any questions• Anyone can do it – limited education

Performance-based codes• Design accepted if performance is met• Requires science and judgement

Performance based codes

Design from firstprinciples

No calculationsprescriptive

Structuraldesign

1

2

3

New Zealandcode structure

Performance based codes

How well has it worked?

Good for a start – few players

Big shift from property to life safety

New entrants (cowboys with computers)

National standards now being set

Need to prescribe design fires

Analysis Fire science

Thermal analysis

Structural analysis

Computing power

Predictive capacity

Computing power

Acc

urac

y

Limits to predictive capacity

Computing power

Acc

urac

yINPUTS:• Fire size• Location• Geometry• Restraint• Materials

– Spalling– Adhesion

• Assumptions

OUTPUTS:• Data overload

• What is failure?

• Knowledge of operator(education)

COMPUTING:

• Micro-macro dilemma

• Trouble-shooting

Micro-macro dilemma

Whole structure

Critical regions

Low resolution

High resolution

OK

Difficult

OKDetail vs complexity

NIST had this problem with WTC

Predictive capacity

Computing power is out-stripping the ability to do improved analysis, because of:

• Limitations on input data• Limited ability to handle output• Lack of large scale test results

Keep it simple.

What can we do now?

Work on problems that can be solved• Computer friendliness• Data management• Material properties• More test data• Thermal analysis• Structural analysis.

What can we do now?

What about the difficult problems ?• Fire size?• Fire location?• Sprinkler reliability?• Changes in use?• Fire after earthquake?• Terrorist attacks?

RISK ASSESSMENT

Risk Assessment

Seismic engineering:1. Determine statistical range of earthquakes2. Subject the building to small, medium,

large earthquakes3. Calculate a distribution of effects4. Estimate likely life-time cost of damage5. Change the design, do it again6. Quantify likely performance

Risk Assessment

Seismic engineering:1. Determine statistical range of earthquakes2. Subject the building to small, medium,

large earthquakes3. Calculate a distribution of effects4. Estimate likely life-time cost of damage5. Change the design, do it again6. Quantify likely performance

Firefires

fires

Must use real fires with decay phase

What are the differences?Earthquakes:• Whole city affected• Seismologists can estimate distribution• Well established design proceduresFires:• When do they happen ?• How severe are they ? • Do the sprinklers work ? • Better design methods needed

Case studies

• Cardington tests• Concrete slabs and steel beams• Concrete walls and steel portal frames• Timber buildings

What can we learn?

Cardington tests

• Good behaviour of unprotected steel frames

Damage to property on the

floor above?

Concrete slabs

Fire tests

SAFIR: In-plane forces

Good analysis, with simple boundary conditions

Red = Compression Blue = Tension

Realistic fires -decay phase temperatures

0

200

400

600

800

1000

1200

0 30 60 90 120 150 180 210 240Time (minutes)

Tem

pera

ture

( o C

)

ISO834 Standard fire with decay phase

Top reinforcing bars

Bottom reinforcing bars

ISO 834 fire

Decay phase

Steel temperatures

Decay temperatures

A1

B1

C1Tensile forces increase as the

fire goes out

Axial force developmentTime

Def

lect

ion

Time

Tem

pera

ture

Time

Stre

ngth

TimeA

xial

forc

e

C

T

With decay phaseTime

Def

lect

ion

Time

Tem

pera

ture

Time

Stre

ngth

TimeA

xial

forc

e

C

T

Tensile forces increase as the

fire goes out

Hollow-core slabs

– Temperature in the voids– Prestressing;

no reinforcement– Effects from surrounding

structural members

Analytical model in SAFIR

long. beam elements

trans. beam el

shell elements

SAFIR analysis

Thermal analysis

Effect of seating details

Correct boundary conditions are critical

for predicting fire behaviour

Boundary conditions

Pin roller

Pin pin

Fixed slide

Fixed fixed

Steel portal frame

Fire behaviour

Difficult to estimate fire severity

SAFIR model

Time = 14.92 minutes

Complex problem:Difficult to give advice to designers

Modular precast concrete

Need to investigate fire performance

Why not do this in wood?

Multi-storey timber buildings

Prestressed timber

What is the fire performance of members, and connections ?

Wood properties in fire

Cone calorimeter tests

Full-scale beam test

Need thermal and mechanical properties of wood

Connection furnace

Bolted connection

Need bearing strength of wood at elevated temps

Intumescent paint

Difficult to calculate intumescent behaviour

Epoxied steel rods

Difficult to calculate epoxy behaviour

Wood-concrete composite floors

Need fire performance of wood, concrete and connections

Conclusions

• Structural fire engineering is challenging.

• Computer analysis is growing fast, but it is not enough on its own.

• We need knowledge about materials, structures, fires, computing, and testing.

• Design and analysis are different skill sets

• Risk assessment will add a new dimension

Thank-you, questions please

Light timber – factory wall

Cannot calculate cracking of gypsum plaster

Timber structures

Cone calorimeter -charring of LVL

Self-drilling dowels

Industrial buildings

What limits predictive capacity?

Computing power

Acc

urac

y

Our computing power is out-stripping our ability to do improved analysis

SAFIR analysis

Time (Minutes)

0 30 60 90 120 150 180

Mid

span

ver

tical

def

lect

ion

(m)

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

ft = 0 MPaft = 1.5 MPaft = 3.0 MPaExperimental results