Research needs towards

Research needs towards resilient communities

Paolo Franchin

2018, June 18th – Thessaloniki, Greece

Community

People Infrastructure

Users System

Engineering

Political, economical, cultural Methods, tools (quantitative)

Obstacles to achieving resilience

Research needs

Resilience

Resilience is difficult to quantify because it is a systemic metric...

0

ft

tR Q t dt

(Bruneau et al 2003)

Sta

te

Time Shock Shock

System

Resilience is difficult to quantify because it is a systemic metric...

…and system is complex

Amplification

Interpolation

(Franchin 2014)(SYNERG, 2009-2013)

Resilience assessment → Resilience design

Assessing resilience is a difficult task, Designing/building it into the system is even more

Two, non mutually exclusive strategies to invest resources

Sta

te

Time

(Bruneau et al 2003)

Pre

ventio

n

Ro

bu

stn

es

s

Redundancy Resourcefulness

Uncertainty in time of

occurrence of shocks

Prevention or cure?

Sta

te

Time

Cure: reasons to wait

Time to next shock

(Meadows et al 2004)

Time to next

election

Prevention or cure?

In praise of prevention

Uncertainty for 𝑡 > 𝑡𝑠ℎ𝑜𝑐𝑘 much larger

What if the shock is TOO LARGE? Preventing the loss may be the only way

𝑡𝑠ℎ𝑜𝑐𝑘

Sta

te =

GD

P

Time

L.A. Northridge 1997

Pre-shock

GDP Disaster

(years)

Pompeii 79

Unsustainable

(never)

Disturbance

Japan, Tohoku 2011 Catastrophe,

partially

unsustainable

(decades)

Progetto CASE, Italy 2009

UNPREDICTABLE

Resilience

Ris

k

(Zio 2018)

(Davis 2014)

Resilience

Research needs

Hazard

Component|Damage

System|Consequences Physical/Function

al

𝑃(𝐷|𝐼)

Intensity 𝐼

Improve

surrogate

models

(fragilities)

As simple as possible,

but not simpler (flow-

based)

Sensitivity to

plug-in models

Simulation at the

base of risk analysis

since the ’70s

because:

• Rare events

• Complex system

• Experimentation

economically

unsustainable &

physically

infeasible

+

Unprecedented

cheap computing

power

Validation

Decision Support

System Using system sims to

inform decision making

𝑃𝑡𝑎𝑟𝑔𝑒𝑡

Resilience-

based

design

(RBD)

1

2

3

1: Resilience-based design

(Bisch 2015-18)

Resilience-based design: 𝑅 → 𝑃

The question is how safe is safe enough?

Codes started with life safety, they’re moving to damage control, will they end up with community resilience?

i.e. Performance-based design with resilience-based targets


System level

Component or sub-system level

Undesired outcome

Core

damage

Radioactivity

release

Reactivity

control Fuel

cooling Confinement

Vital plant functions

Frontline systems

Containment

spray system

Support systems

Water Electrical

System level

Component or sub-system level

TO

P-D

OW

N

BO

TT

OM

-UP

Outmigration

Housing Education Employment

Vital community functions

Undesired outcome

Primary systems (bldgs)

Residential Offices Factories Schools Hospitals

Secondary systems (lifelines)

Water Waste

Public

services

Energy Commun

ication Transport

Unsustainable

impact

Nuclear Power Plant Community

(Mieler et al 2013)


Establish VCF event

trees

Step 3

Combine VCF trees

into single community

tree

Step 4

Com

munity

resili

ence g

oal

p

iji , jå £ Pmax

NO

Primary and

secondary systems

Shock 25 15

C >25%25%>C >15%

C <15% 0.6p

ij

Public services:

tracking variable =

Percent capacity disrupted C

0.3

0.1

Shock 20 10

R >20%20%> R >10%

R <10% 0.7p

ij

Housing:

tracking variable =

Percent residents displaced R

0.2

0.1

Establish acceptable

probability of

undesired outcome

Step 2

Pmax

Establish community

undesired outcome

Step 1

e.g.outmigration

Com

munity

event

tree

Evaluate

mean annual target for

each VCF tracking

variables

Step 6

YES

E Réééé= Rjp1 jj=1

3

é

E Céééé= Cjp2 jj=1

3

é

Expected annual disruption compatible

with community resilience goal

20 10

Housing

(R)

Public

services (C)

25 15 0.42

0.21

0.070.12

0.06

0.02

0.06

0.03

0.01

R <10%&C <15%

R >20%&C >25%

0.7´0.6=0.42

0.1´0.1=0.01

0.42

0.210.07

0.12

0.06

0.02

0.06

0.03

0.01

Identify

adverse event

sequences causing

undesired outcome

and associated

probabilities

Step 5

20 10

Housing

(R) Public

services (C)

25 15

p

iji , j

adverse

å =0.25

(Mieler et al 2013) ?


Systemic analysis can fill this gap

{{

1 1

VCF service system system

n n m m

D I I D1 2 3

Public services

Service

importance

matrix (1×n)

serviceI

Service P

olic

e

Health-c

are

Fo

od

0.3 0.3 0.4

Serv

ices

0.4

Systems

Police

Health-care

Food

Primary Secondary

Polic

e s

tatio

ns

Hospitals

Sto

res

Wate

r

Ele

ctr

ic p

ow

er

Fu

el (o

il)

Roads

0.5

0.5

0.2 0.20.2

0.2 0.10.10.1

0.1 0.10.20.1

System

importance

matrix (n×m)

systemI

New hospital tolerable

disruption (unknown)

Disruption to existing

Buildings and lifelines

(from systemic analysis!)

,2 ,2

,1

,1

VCF service system system

system

service system

DD

I I D

I I

Still

just an

idea

{ {

,1 ,2 ,1

,2

,1 ,1 ,2 ,2

1 11 1 1 ( 1) ( 1) 1

1 1 1 1

VCF service

system system system

system

service system system service system system

n nn n m m

DD

D

II I

D

I I I I D1 2 3 14 2 43 1 2 3 14 2 43

1 4 2 4 3 1 4 4 4 2 4 4 43

Resilience-based performance target for a new hospital

Hospital ∈ Health-Care System ∈ Public Services VCF

𝑃𝐶 = 2 × 10−4 ÷10−5?

2: System model

*

* *

T

D D seismic D

S S D D

I q Q h Q h 0

I h I h r q 0

% %% % %%

%% %% % %

o

*

* *

with

T

D

S S D D

I q Q 0

h r q I h I h r q 0 r q Rq q

System functional model

0 1 1 0

0 1 0 1

1 0 0 1

1 0 1 0

0 0 1 1

e vn nI

e3

e4

e5

e2

e1 v4

v1

v3

v2

4

5

V

E

n

n

e3

e4

e5

e2

e1 v4

v1

v3

v2

0 11 0

0 1 0 1* ** 1 0 0 1

1 0 1 0

0 0 1 1

e ve De S

n n n nn nS DI II

e3

e4

e5

e2

e1 v4 demand

v1 source

v3

v2

V S Dn n n

1 0 1 1

0 1 1 1

1 1 1 1

1 1 1 1

v vn nA

Head

driven

Additional

demands

Balance (flow continuity at nodes)

Resistance (line loss)

1

1

edge flows

node demands

Q 0 if junction

e

D

n

n

i

q

Q

Assessment problem,

not design! • Post-event demand model

must be linked to systemic

damage

• Done for water and gas

• Source capacity still missing

Connectivity Flow

(Cavalieri et al 2014)

System functional model

Power networks: much more difficult problem (SECD formulated in 1955 still no fast/robust solution technique)

People are doing everything ranging from pure connectivity, to DC (linearized), to AC (nonlinear). Truth is that even AC is incomplete

Power networks

3: Components’ models

Systemic analysis → 100s or 1000s of components → surrogate models

Component damage model

𝑝 𝐿𝑆𝑖,2 𝐼𝑖

𝑝 𝐿𝑆𝑖,1 𝐼𝑖

𝐼𝑖 = 𝑥

Intact

Damaged

Collapsed

𝑝 𝐶𝑖 = 0 𝑥



𝐼𝑖 is just one parameter of ground motion

𝑰𝑖|𝐼𝑖 other GM parameters depend on site

Fragility is structure & site-dependent

Fragility from field damage → difficult to generalize → numerical

simulation

𝐼𝑖 𝐶𝑖

fragility 𝑝(𝐿𝑆𝑖𝑗|𝐼𝑖) → damage given intensity 𝑝(𝐶𝑖|𝐼𝑖)

Calibrated models


Fragility analysis via numerical simulation is a delicate business. Results depend on: ground motions, numerical model, analysis

method, statistical method and modelled uncertainty

Refined fragility analysis of archetype buildings should not be used to support fragility functions for classes of assets!


(Borzi et al 2015)(Franchin et al 2016)(RINTC Workgroup, 2018)

0 0.1 0.2 0.3 0.4 0.50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PGA (g)

Da

ma

ge

fra

gili

ty

0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PGA (g)

Colla

pse

fra

gili

ty

0

20

40

60

80

100

120

140

160

180

0

10

20

30

40

50

−25

−20

−15

−10

−5

0

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.40

1

2

3

4

5

PGA (g)

Y

bridge database (a) bridge information model (b)

bridge finite element model (c) multiple stripe analysis (d)

Carriageway (3 decks)!

Deck (1 span)!

Deck (3 spans)!

Deck (1 span)!

Pier segments!

Foundation!

Deck joint!

Abutment!

Deck (3 spans)!

Spans!

Span segments!

Individual runs (each is one two-components

ground motion)"

Lognormal fit

to analysis results"

Deck: Elastic frame elements"

Restrained node"

Free node"

Link/Connection:

twoNodeLink element"

Fixed-base (double-node, option for SSI & multiple-support excitation)"

Base nodes of pier segment

constrained to upper node of foundation"

Deck: intermediate masses/dofs"

Deck joint: zeroLength element"

Pier segments & transverse beams:"

BeamWithHinges element"

D4 project 2009-2012 450+ bridges

𝐼𝑖 𝐶𝑖

𝑀

𝜂 𝜀 𝐸

𝐵𝐺

Bridge

geometry

𝐵𝑀

Bridge

materials

𝑆475

Site seismicity:

fragility is site-dependent

BN-based fragility

Lo

ss

direct 𝐿𝐷 = ∑𝐿𝐷𝑖 indirect 𝐿𝐼 = ∑𝐿𝐼𝑖 + ∑∑𝐿𝐼𝑖𝑗

RINTC project 2015-2018 Tens of RC, PRC, URM, steel

buildings

Conclusions

• Resilience is improved by reducing vulnerability and improving response/recovery

• Vulnerability reduction seems the most reliable, given the uncertainty in 𝑡 > 𝑡𝑠ℎ𝑜𝑐𝑘

• Components’ damage: need better surrogate models Fundamental research in structural and geotechnical engineering is still needed

• Systems’ behaviour: need more realistic representation (flow! Or enhanced/smart connectivity…)

• If former two are achieved, systemic analysis will be reliable enough to link performance of the components to global community resilience goals. This will provide: • A rational basis for performance targets in next generation

codes

• Support for building decision-support systems for use in real time

Conclusions

Thank you!

Funding:

European Commission:

FP7 project SYNER-G – Systemic analysis framework

Italian Department of Civil Protection:

Reluis project RINTC – Seismic risk of Italian code-conforming buildings

Reluis project RS6 – Seismic risk to lifelines

EUCENTRE project d4 – Seismic vulnerability of Italian highway bridges

Contributors:

Francesco Cavalieri, Pierre Gehl G. Weatherill, F. Noto, A. Lupoi, F. Mollaioli, S. Tesfamariam, S. Giovinazzi

Documents

Research needs towards