Assessment of Critical Infrastructure System Interdependencies

1
Assessment of Critical Infrastructure System
Interdependencies
PHD Dissertation Proposal

• Ping Chen
• Department of Civil and Environmental Engineering
• Carnegie Mellon University
• 2005-09-18

2
Motivation

• Infrastructure Systems
• Function collaboratively and synergistically
• Produce/distribute a continuous flow of essential
  goods/services
• Interact with peoples daily life and everyday
  economic activities.
• Information and communications, physical
  distribution, energy, banking and finance, and
  vital human services PCCIP 2003
• Pros and Cons of Rapidly Growing Interdependence
• Enable the provision of advanced, efficient and
  convenient service
• More vulnerable than before interdependence-induc
  ed threat
• Potential threats come from multiple facets
  inside or outside natural or manmade
• Further efficiencies might be difficult to
  realize because of tradeoffs with induced
  vulnerabilities Heller 2002

3
Research Questions

• Given the current status of knowledge about
  infrastructure interdependence which is confined
  to the realm of qualitative analysis, what would
  be effective and feasible measures that can
  quantitatively represent their interdependencies?
  
• What kind of model is able to estimate the
  interdependency using the measures defined above
  and how does one build it?
• How does one validate whether the proposed model
  is effective for representing the interdependency
  effect and what are the uncertainties in this
  model?
• How could the derived model inform us about the
  vulnerability of individual infrastructure
  sectors?

4
Overview of Approach and Defined Tasks

• Task 1 Review Current Knowledge on
  Infrastructure Interdependence and Propose
  Quantitative Measurement Strategy
• Task 2 Review and Evaluate Interdependence
  Modeling Methods
• Task 3 Develop Resilient Supply Driven
  Infrastructure Interdependence Modeling Method
• Task 4 Apply and Interpret Revised
  Infrastructure Interdependence Model
• Task 5 Analyze Model Uncertainty and Verify
  Model Soundness
• Task 6 Assess Interdependence Induced Critical
  Infrastructure Vulnerability

5
Task 1 Current State of Knowledge on
Infrastructure Interdependence
Examples of Infrastructure Interdependence
Six Dimensions of Infrastructure Interdependence
Source Rinaldi et al, Identifying,
Understanding, and Analyzing Critical
Infrastructure Interdependence
6
Proposed Interdependence Measurement Strategy

• Consequence Measurement vs. Dimensional
  Measurement
• Indicators (1) economic loss (2) inoperability
  Haimes 2001
• Identified major connection type Physical
  interconnections through supply chain linkages
• Total consequences measurement (direct
  indirect)
• Two dependence-related disruption type

Type I Demand-driven
Type II Supply-driven
Direction of Physical Connection, from Supply
Sector to Demand Sector
Interconnected Infrastructure Sector
Initially Disrupted Sector
7
Task 2 Review and Evaluation of Relevant
Modeling Methods
Industry-to-Industry Transaction Data in Detail

• Leontief IO Model

• Ghosh IO Model

8
Effects of Generating 1 million Final Demand in
Power Generation Sector
Effects of Providing 1 million Primary Input in
Power Generation Sector

• Ghosh IO Model
• v (0, 0, ,0, 1, 0,0)

• Leontief IO Model
• c (0, 0, ,0, 1, 0,0)

9
Current Status of Extending EIO Model to
Demand-Driven Infrastructure Interdependence
Modeling

• c - the demand-side perturbation vector
• A - the interdependency matrix
• q - the accumulated inoperability vector

Telecommunication
Transportation
Water
Power
Power
Supposed Interruption in the transportation
sector by 10
Transportation
Example
Water
Telecommunication
Haimes and Jiang 2001 Haimes, Y. Y., Jiang,
P. Leontief-based model of risk in complex
interconnected infrastructures, (2001)
10
Extending Ghosh Model to Supply-Driven
Interdependence Model?
Ghosh Model Reexamination Power Generation
Perturbation
11
Extending Ghosh Model to Supply-Driven
Interdependence Model? (cont)

• Difficulty and Limitation?
• Less applications of supply-driven model than
  demand-driven model
• Tracing forward impact is harder than backward
  impact substitution, backup ability, etc
• How to consider pre-planned risk mitigation
  actions
• More Constraints on applying Ghosh-based Supply
  Driven Model
• Critical supply sectors bottleneck effect
• Various effects from different levels
  Interdependence
• The worst consequences of cascading disruption
  under constrained supplies

12
Infrastructure Dependence Survey

• Objective
• Supplement the information from the economic
  dependence data
• Acquire professional assessment on
  interdependence
• Consider the practical experiences
• Surveyed Sectors
• Water Plant (PA and AZ)
• Power Plant (PA)
• Response Rate
• Water Plant (16/64 25)
• Power Plant (1/20 5)

13
Infrastructure Dependence Survey (cont)

• Sample Question Assume that you have your
  normal operation at your water supply plant, but
  that the input from one of these infrastructure
  sectors is completely cut off, due to a strike,
  accident or other disruptions. What is your
  estimation of the longest time that your normal
  water supply could continue based only on your
  normal inventory or back up (e.g. by using your
  own electric generation capacity) with a
  disruption from one of the following
  infrastructure sectors?

4
3
2
1
Point
5
14
Interdependence Evaluation from Practitioners
Box Plot of Survey Responses (median, 1st
quarter, 3rd quarter)
95 Confidence Interval (mean) Plot of Survey
Response
15
Interdependence Evaluation from Practitioners
Responded Impact Assessment vs. Direct Requirement
Responded Impact Assessment vs. Direct Supply
16
Summary from Evaluations of Existing
Interdependence Model

• EIO model, both Leontief and Ghosh model provide
  good bases for extending to infrastructure
  interdependence model
• Meanwhile, both models are criticized for the
  assumption of fixed coefficients, supply driven
  model is suffered more than the demand driven
  model because of bottleneck effect
• The distribution of the survey data is close to
  the output of the supply-driven model rather than
  the demand-driven model
• The survey data provides a different information
  source other than the economic interconnections
  and it evaluates the sectors empirical ability
  on offsetting the negative effect of supply
  unavailability
• For measuring the all-around impact that is
  proposed, the survey data can supplement the
  economic input output data to form a more
  resilient supply driven infrastructure
  interdependence model

17
Task 3 Develop Resilient Supply-Constrained
Infrastructure Interdependence Modeling Method

• Phase I Establish Industry-to-Industry Direct
  Supply Matrix
• Phase II Formulate Supply Constrained (SC)
  Economic Interdependence Matrix Improved by
  Empirical Evaluation
• Phase III Derive Supply Constrained Total Impact
  Estimation Model

A Direct Requirement Matrix the vector of
output from each sector
B Direct Supply Matrix S Empirical
Interdependence Matrix
(1) Economic Loss Impact
(2) Inoperability Impact
18
Phase II Revised Supply Constrained
Interdependence Model (SCIM)

• Method 1 Simple Weight Updated SCIM Multiplier
• Method 2 Bayesian Method Updated SCIM Multiplier
  
• Method 3 Empirical Knowledge Based Inoperability
  Estimation SCIM Multiplier

19
Task 4 Apply and Interpret Revised
Infrastructure Interdependence Model

• Scenario 1 General Inter-infrastructure Impact
• Estimate the chain disruptions among
  infrastructure sectors and non-infrastructure
  sectors for hypothetically generated
  perturbations.
• Scenario 2 Energy Security Concern
• Estimate the influence on the entire economic
  sectors as the result of reduced amount of
  imported gasoline
• Relevant energy security cases need to be further
  collected and adapted to fit the data requirement
  of the proposed model
• Scenario 3 Transportation Access Connectivity
• Apply the decomposed SCIM on regional
  transportation interruption events, such as labor
  strike
• Compare the impact from curtailed transportation
  accesses on adjacent states
• Potential data sources
• Transportation Satellite Account (TSA)
• Regional I-O Multiplier System (RIMS II)

20
Task 5 Analyze Model Uncertainty and Verify
Model Soundness

• Interdependence Model Uncertainty
• Survey uncertainty
• EIO-LCA Uncertainty Analysis Experience
• Old data
• Incomplete data
• Aggregation
• Imports
• Validation Case Collection and Validation Process
• Validation Case Requirement
• Report of economic loss of infrastructure sector
  in continuous days during disruption events
• EIO-LCA Validation Experience
• Comparison of the relative impact on different
  sectors would be more reliable and representative
  than comparison of absolute impact with the real
  occurrence

21
Task 6 Assess Interdependence Induced Critical
Infrastructure Vulnerability

• Dependence induced vulnerability sources from
  both demand-driven disruption threats and
  supply-constraint disruption threats
• Different order of independence offer different
  effect on vulnerability
• Two influential factors that are identified to
  assess vulnerability
• The probability of initializing disruptions on
  various sectors
• The total impact of the implementation of initial
  disruptions on the sector of interest directly
  and indirectly
• Proposed vulnerability measurement
• Sensitivity Analysis on Pi and Rm

Rm The vulnerability assessed on infrastructure
sector m (m1,..,n) Pi The probability for the
formation of initial disruption in source i Cim
The consequence of the occurrence of an initial
disruption from source i on the assessed
sector m n The number of infrastructure
sectors
22
Expected Contributions

• Consequence measurement has been proposed that
  can feasibly and effectively measure
  infrastructure interdependence
• Demand-driven and supply-constrained interruption
  types are used to categorize various supply chain
  related interruption occurrences
• Empirical Supply-Driven Infrastructure
  Interdependence Model (EMSD-IIM) has been
  developed that can assess both direct and
  indirect disruption impacts
• The empirical evaluation of interdependences has
  been, for the first time, quantitatively
  incorporated to count the risk-mitigation
  capacity of these sectors in assessing
  interdependence effect
• The proposed interdependence induced
  vulnerability measurement considers the
  occurrence of two interruption types. In
  addition, the contribution to the overall
  vulnerability from various level interdependences
  can be adjusted

23
TimeLine
24

• Questions, Comments ??

25
Motivation

• Infrastructure Systems
• Infrastructure systems are generally manmade
  systems that function collaboratively and
  synergistically to produce and distribute a
  continuous flow of essential goods and services
  that interact with peoples daily life and
  everyday economic activities. Heller 2002
• Information and communications, physical
  distribution, energy, banking and finance, and
  vital human services PCCIP 2003
• Infrastructure Interdependence Histories
• 1999 June, WA, SCADA system failure, Pipeline
  system shutdown for one year
• 2001 Early, CA, electricity outage, total loss
  over 1billion
• 2001 July 19, Baltimore Tunnel Train Derailment
  and Fire
• 2001 Sep 11, NY WTC Terrorist Attack
• Pros and Cons of Rapidly Growing Interdependence
• Enable these infrastructure sectors to provide
  more advanced, efficient and convenient service
  for use
• Threats on infrastructure sectors come from
  multiple facets inside or outside natural or
  manmade, etc.
• Infrastructures now become more vulnerable than
  before because of interdependence-induced threat
• Further efficiencies might be difficult to
  realize because of tradeoffs with induced
  vulnerabilities Heller 2002

26

• Leontief IO Model
• c (0, 0, ,0, 1, 0,0)
• output
• Application
• Energy Consumption Estimation
• Environmental Implication (www.eiolca.net)

• Ghosh IO Model
• v (0, 0, ,0, 1, 0,0)
• output
• Application
• International Comparisons, Forward Linkages, and
  Key Sectors
• Sectoral Impact Studies

Final Demand Power Generation and Supply
(million)
Primary Input Power Generation and Supply
(million)
27
Task 3 Develop Resilient Supply-Constrained
Infrastructure Interdependence Modeling Method

• Phase I Establish Industry-to-Industry Direct
  Supply Matrix
• Phase II Formulate Supply Constrained (SC)
  Economic Interdependence Matrix Improved by
  Empirical Evaluation
• Phase III Derive Supply Constrained Total Impact
  Estimation Model

A Direct Requirement Matrix the vector of
output from each sector
B Direct Supply Matrix S Empirical
Interdependence Matrix
(1) Economic Loss Impact
(2) Inoperability Impact
28
Output identity
Final Demand
Fixed Input Coefficients

• Leontief IO Model

• Ghosh IO Model

Primary Input
Fixed Output Coefficients
Input identity
29
Preliminary Result and Analysis (Simple Weight
Method)
1. Direct Supply Matrix
2. Survey Response
Dependence Multiplier (Direct Supply vs. Adjusted
Direct Supply)
Survey Response Normalization
3. Simple Weight Updated Coefficients ( )
30
Preliminary Result and Analysis (Simple Weight
Method) (cont)
31
Review and Evaluation of Relevant Modeling
Methods (cont)
Output identity
Final Demand
Fixed Input Coefficients

• Leontief IO Model

• Ghosh IO Model

Primary Input
Fixed Output Coefficients
Input identity
32
Extending Ghosh Model to Supply-Driven
Interdependence Model?

• Feasibility?

Example Perturbation Sector Power Generation
and Supply Perturbation Scale Inoperability
10