Network Impact Assessments Under Unscheduled Events

1
Network Impact Assessments Under Unscheduled
Events
The MAE Center Student Leadership Council
Seminar
The Social Impact of Earthquake
Engineering Policy Making

• May 6, 2004
• Tschangho John Kim
• Endowed Professor of Urban and Regional Systems

2
Economic Impact of EQ

• The New Madrid Seismic Zone, located in the heart
  of the United States have great impacts on
  economic activities related to this and
  surrounding regions.
• Based on the Commodity Flow Survey (1993) by the
  Bureau of Transportation Statistics, about 1/3 of
  total commodity flows in the US are related to
  the Midwest, including this New Madrid Seismic
  Zone.
• If a catastrophic earthquake occurs in this area,
  the direct and indirect economic impacts will
  spread far beyond the region, and will have
  sizable impacts in other regions.

3
Project Goals and Objectives

• Estimating economic losses due to network damage,
  taking into account time sequencing and
  different, but a flexible spatial configuration.
• Temporal division will be divided into four
  quarters per year.
• With flexible regional specification, the
  economic impacts of events located in different
  parts of the region could be generated with
  greater attention to detail in areas closer to
  the event.
• This research will expand the GIS system
  currently in progress to develop capabilities for
  a greater degree of flexibility in spatial
  assignment

4
DATA DESCRIPTION Data Source

• 1997 Commodity Flow Survey (CFS) Data by the
  Bureau of Census, US Department of Commerce
  (Total Annual National Freight Cost 6,944
  Billion)
• 1997 National Transportation Atlas Database by
  U.S. Department of Transportation
• Use of the US Interstate Highway Network
  constructed with 167 nodes and 532 links
• Use a spider or spanning railway network which is
  connected by each pair of adjacent subregions

5
Overview of the Analysis
Run Transportation Network Model
No Earthquake
Stochastic Formation
Location Magnitude Year
Run Transportation Network Model
Network Loss Function
Earthquake
Network Disruption Ratio
Economic Sector Weights
Final Demand Loss Function
Resiliency Factor
I-O Table
Regional Economic Structure
Transportation Network Engineering
Interindustry Impact
Potential Benefit (unit dollar-mile)
B-C Analysis
Retrofit Cost Function for Corresponding Network
Components
Potential Cost (unit dollar)
6
83 EQAZ Map
7
Highway Network
8
Railway Network
9
13 Economic Sectors

• Agriculture, Forestry, and Fisheries
• Mining
• Construction
• Food and Kindred Products
• Chemicals and Allied Products
• Primary Metals Industries
• Fabricated Metal Products
• Industrial Machinery and Equipment
• Electronic and Electric Equipment
• Transportation Equipment
• Other Non-Durable Manufacturing
• Other Durable Manufacturing
• TCU, Services, and Government Enterprises

10
Model Formulation

• Integrating the Standard Interregional Flow Model
  with a Transportation Network Model
• A Development of a Network-Based Commodity Model
  based on Wilson(1970), and Leontief and
  Strout(1963)
• The proposed model minimizes the total shipment
  cost subjected to several constraints

11
Model Performance
12
Scenario Analysis

• Assume that the highway link (section) has been
  disconnected by an earthquake
• Scenario A between Chicago, IL and Gary, IN on
  I-94
• Scenario B between Louisville, KY and Nashville,
  TN on I-65
• Scenario C between Little Rock, AR and
  Nashville, TN on I-40
• Scenario D Scenario A Scenario B
• Scenario E Scenario A Scenario B Scenario C
• Compare the travel costs and the travel lengths
  from the model for both the case with the link
  and the case without the link

13
Scenario A
Scenario D B C
Scenario E A B C
Scenario B
Scenario C
14
Results of Scenario Analysis
Percentage changes of objective function value
(total transportation cost)
Scenario A Total Cost 0.31
Scenario D (BC) Total cost 0.44
Scenario B Total cost 0.22
Scenario E (ABC) Total cost 0.75
Scenario C Total cost 0.25
15
Result of Scenario Analysis
Changes of Objective Function Value (Total
Transportation Cost) and Mean Shipping Length
16
Result of Scenario Analysis

• I-94 considered in Scenario A involves more
  economic activities than I-65 in Scenario B and
  I-40 in Scenario C even if the Scenario A is
  further from New Madrid (epicenter) than Scenario
  B and C.
• A total cost change 0.31
• B total cost change 0.22
• C total cost change 0.25

17
Result of Scenario Analysis
Changes of Commodity Flows for Scenarios (
billion)
18
Result of Scenario Analysis

• In Scenario A, the interregional commodity flows
  as much as 7.21 billions are transferred to the
  intraregional commodity flows because of
  increasing the shipping costs in highway due to
  the disruption of the highway network section.
• In addition, the total OD commodity flows by the
  highway mode are converted as much as 3.13
  billions to the railway mode.

19
Stand Alone Application

• Calculating the damage of the bridges by using
  the fragility curves
• The data of bridges extracted from National
  Bridge Inventory (NBI) for Mid-Eastern area.
• The application allows the users to pick their
  earthquake scenario consisted of the location of
  epicenter and the moment of magnitute.

20
(No Transcript)
21
Improvements
In progress
22
Web Application

• Developing the web application instead of the
  stand alone application for interoperability
• Via the web, the users can run the model and
  explore the results
• Improving visualization
• Visualizing regional economic situation before an
  earthquake
• Visualizing economic loss after an earthquake
  effectively
• Visualizing disrupted links and bridges
• Improving scenario management
• User can create and manage own earthquake
  scenario,
• Storing the result of running the model
• Exploring the result by scenario

23
Web Application System Architecture
24
ltlt Scenario management gtgt
25
ltlt Exploring the baseline data (before an
earthquake) gtgt
26
Most-Likely Path Flow (MLPF)

• The solution of link flow is unique and it can be
  proven, however in general the solution of path
  flow is not unique - the path flow has multiple
  solutions.
• We assume that the best solution is the most
  likely path flow that is occurred in the highest
  possibility among the multiple solutions.
• It will allow to know which flow passes through
  or terminates at the zone.

27
Windowing

• Currently flows and damages are obtained for a
  compact national network of highways.
• If a state is interested in estimating damages
  and minimizing the losses within the state, it
  will need to have more specific information on
  commodity flows involving more zones and highways
  in the state.
• Through windowing, we want to utilize the results
  obtained from the current approach to find the
  commodity flows inside a region or county within
  a state in order to assess damages at local
  level.
• Disaggregation techniques are required to solve
  for the local level using solutions obtained from
  the current approach.

28
(No Transcript)
29
Spatio-Temporal Data Model for Unscheduled Events
(1)

• Feature based Approach
• Since frequency of update is not often, the
  time is not major concern
• However, the attribute of feature is major
  concern of this project.
• Also, various feature (line, points, and
  polygons) should be handled
• Therefore, the feature based approach is suitable
  for this project.
• Among the models in feature based approach
• Space-Time composite model (Langran 1992 )
• For transportaion data and economic data
• Spatio-temporal object model (Worboys 1994a
  1994b)
• For the collection of epicenter data

30
Spatio-Temporal Data Model for Unscheduled Events
(2)
31
Spatio-Temporal Data Model for Unscheduled Events
(3)

• Model contains 3 components
• Space-Time Composite (STC) components
• Snap Shot (SS) components
• Spatio-Temporal Object (STO) components
• STC components are built from SS components (Shaw
  and Xin 2003)

32
Spatio-Temporal Data Model for Unscheduled Events
(4)

• Related ISO standards

33
Integrated Commodity Flow Model A Snapshot Model
ICFM
FDLM
34

• Exogenous variables

• Endogenous variables

35
Integrated Commodity Flow Model Conceptual Scheme

• if the user would like to explore the
  transportation flow and multi-regional input
  output analysis after n time intervals (at tn)
  since the earthquake occurred, the iterations
  might last until time tn.

36
Future Study

• Implementing Spatio-temporal database
• DMBS PostgreSQL 7.4.1
• Spatial extension PostGIS 0.8.1
• Implementing Spatio-Temporal Network Economic
  Loss Models