A Vision for Regional PBSA of Transportation Networks

Transcription

1 A Vision for Regional PBSA of Transportation Networks by Ertugrul Taciroglu, Professor Civil & Environmental Engineering Department University of California, Los Angeles Faculty Affiliate, UCLA

2 A Vision for Regional PBSA of Transportation Networks + work in progress by Ertugrul Taciroglu, Professor Civil & Environmental Engineering Department University of California, Los Angeles Faculty Affiliate, UCLA

3 A Vision for Regional PBSA of Transportation Networks + work in progress

4 A Vision for Regional PBSA of Transportation Networks + work in progress

5 A Vision for Regional PBSA of Transportation Networks + work in progress

6 Outline B. John Garrick Institute of Risk Sciences

7 Outline Motivation and objectives B. John Garrick Institute of Risk Sciences

8 Outline Motivation and objectives Vision and scope B. John Garrick Institute of Risk Sciences

9 Outline Motivation and objectives Vision and scope Details of envisioned components B. John Garrick Institute of Risk Sciences

10 Outline Motivation and objectives Vision and scope Details of envisioned components Some preliminary results and outlook B. John Garrick Institute of Risk Sciences

11 Motivation and Objectives

12 Why regional assessment?

13 Why regional assessment?

14 Why regional assessment?

15 Why regional assessment? Hazards affect regions. The big picture is needed for Actuarial plans (insurance companies) Urban planning & public policy (government) Emergency service planning (1 st responders)

16 Why regional assessment? Hazards affect regions. The big picture is needed for Actuarial plans (insurance companies) Urban planning & public policy (government) Emergency service planning (1 st responders) Built environment is highly interconnected Residential neighborhoods, business centers Transportation networks Lifelines (water, power, communications)

17 Why regional assessment? Hazards affect regions. The big picture is needed for Actuarial plans (insurance companies) Urban planning & public policy (government) Emergency service planning (1 st responders) Built environment is highly interconnected Residential neighborhoods, business centers Transportation networks Lifelines (water, power, communications)

18 Why regional assessment? Hazards affect regions. The big picture is needed for Actuarial plans (insurance companies) Urban planning & public policy (government) Emergency service planning (1 st responders) Built environment is highly interconnected Residential neighborhoods, business centers Transportation networks Lifelines (water, power, communications)

19 Challenges

20 Challenges Data metadata models

21 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools)

22 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools) Access to detailed data may be not be possible (requires estimation missing data, machine learning)

23 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools) Access to detailed data may be not be possible (requires estimation missing data, machine learning) Processing requires large computational resources (would break records for civil engineers)

24 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools) Access to detailed data may be not be possible (requires estimation missing data, machine learning) Processing requires large computational resources (would break records for civil engineers) Models decision variables

25 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools) Access to detailed data may be not be possible (requires estimation missing data, machine learning) Processing requires large computational resources (would break records for civil engineers) Models decision variables Heterogeneous analysis tools (OpenSees, OpenSHA, PACT)

26 Challenges Data metadata models Diverse sample population (requires sophisticated and as of yet non-existent data harvesting tools) Access to detailed data may be not be possible (requires estimation missing data, machine learning) Processing requires large computational resources (would break records for civil engineers) Models decision variables Heterogeneous analysis tools (OpenSees, OpenSHA, PACT) New tech needs to be brought in (data analytics, Bayesian inference, etc.)

27 Objectives Risk framework for a highway network (Miller & Baker, 2015)

28 Objectives Develop a (semi-) automated interactive platform that can evaluate seismic vulnerability of complex transportation networks: Risk framework for a highway network (Miller & Baker, 2015)

29 Objectives Develop a (semi-) automated interactive platform that can evaluate seismic vulnerability of complex transportation networks: 1. Generate structural models using data harvested from various sources Risk framework for a highway network (Miller & Baker, 2015)

30 Objectives Develop a (semi-) automated interactive platform that can evaluate seismic vulnerability of complex transportation networks: 1. Generate structural models using data harvested from various sources 2. Carry out site- and structurespecific seismic analyses Risk framework for a highway network (Miller & Baker, 2015)

31 Objectives Risk framework for a highway network (Miller & Baker, 2015) Develop a (semi-) automated interactive platform that can evaluate seismic vulnerability of complex transportation networks: 1. Generate structural models using data harvested from various sources 2. Carry out site- and structurespecific seismic analyses 3. Evaluate the consequent economic losses at the network-level

32 Objectives Existing predictive computational tools and IT capabilities allow unprecedented granularity for such seismic risk and loss assessment studies Risk framework for a highway network (Miller & Baker, 2015)

33 Objectives Existing predictive computational tools and IT capabilities allow unprecedented granularity for such seismic risk and loss assessment studies Risk framework for a highway network (Miller & Baker, 2015)

34 Objectives Risk framework for a highway network (Miller & Baker, 2015)

35 Objectives Hasn t been done before [at site-, structure-, and scenario-specific granularity] Risk framework for a highway network (Miller & Baker, 2015)

36

37

38 Early media reports estimated losses around $4B (based on a insurance agency report)

39 Early media reports estimated losses around $4B (based on a insurance agency report)

40 Early media reports estimated losses around $4B (based on a insurance agency report) One month after the event, the total losses were reported to be ~$400M (Kale, 2014)

41 Vision and Scope

42 Regional PBSA of Ordinary Caltrans Bridges

43 Regional PBSA of Ordinary Caltrans Bridges

44 Regional PBSA of Ordinary Caltrans Bridges

45 Regional PBSA of Ordinary Caltrans Bridges

46 Regional PBSA of Ordinary Caltrans Bridges

47 Regional PBSA of Ordinary Caltrans Bridges

48 Regional PBSA of Ordinary Caltrans Bridges

49 Regional PBSA of Ordinary Caltrans Bridges

50 Regional PBSA of Ordinary Caltrans Bridges

51 Regional PBSA of Ordinary Caltrans Bridges

52 Regional PBSA of Ordinary Caltrans Bridges

53 Regional PBSA of Ordinary Caltrans Bridges

54 Regional PBSA of Ordinary Caltrans Bridges

55 Details of the Envisioned Components

56 Regional PBSA of Ordinary Caltrans Bridges

57 Regional Data PBSA of Ordinary Caltrans Bridges

58 Where will the data come from?

59 Where will the data come from? National Bridge Inventory (NBI) by FHWA

60 Where will the data come from? National Bridge Inventory (NBI) by FHWA Caltrans Bridge Database

61 Where will the data come from? National Bridge Inventory (NBI) by FHWA Caltrans Bridge Database California Strong Motion Instrumentation Program (CSMIP) Database

62 Where will the data come from? Guideline Documents

63 Where will the data come from? Guideline Documents Caltrans Standard Plans allow determination of many metadata elements (e.g., abutment seat length, shear-key reinforcement, foundation configuration, etc.) Caltrans Seismic Design Criteria Manual (Caltrans SDC) provides era-specific information on component and system design

64 Where will the data come from? Internet Harvesting

65 Where will the data come from? Internet Harvesting Google Maps/Earth, MapQuest, etc. can be interrogated online more on this later

66 Where will the data come from? Internet Harvesting

67 Where will the data come from? Internet Harvesting Crowd Sourcing - uses human intelligence when algorithms are too difficult to devise - wikipedia-type consensus models can be built (contributors v. referees)

68 Where will the data come from? Internet Harvesting Crowd Sourcing - uses human intelligence when algorithms are too difficult to devise - wikipedia-type consensus models can be built (contributors v. referees) Typical Seat Abutment

69 Where will the data come from? Internet Harvesting Crowd Sourcing - uses human intelligence when algorithms are too difficult to devise - wikipedia-type consensus models can be built (contributors v. referees) Typical Seat Abutment typical Is this a seat type abutment (Yes/No)?

70 Image to Model Detection of Bridge Locations

71 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI

72 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI Extract a satellite image of the location corresponding to approximate bridge coordinate

73 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI Extract a satellite image of the location corresponding to approximate bridge coordinate Run a road extraction algorithm to detect roads on the selected image

74 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI Extract a satellite image of the location corresponding to approximate bridge coordinate Run a road extraction algorithm to detect roads on the selected image Generate random points on detected continuous road lines and pass coordinates to Google Roads API

75 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI Extract a satellite image of the location corresponding to approximate bridge coordinate Run a road extraction algorithm to detect roads on the selected image Cross-check route inventory with NBI, then highlight the relevant road line Generate random points on detected continuous road lines and pass coordinates to Google Roads API

76 Image to Model Detection of Bridge Locations Read approximate bridge coordinates from NBI Extract a satellite image of the location corresponding to approximate bridge coordinate Run a road extraction algorithm to detect roads on the selected image Prompt user to mark beginning and end points Cross-check route inventory with NBI, then highlight the relevant road line Generate random points on detected continuous road lines and pass coordinates to Google Roads API

77 Image to Model Developing the Wireframe Bridge Models Create up to 1000 sampling points between user-defined coords. Snap points to road centerline curve using Determine ground elevations using Determine road elevations using Establish wireframe bridge model Create virtual cameras from bridge centerline curve Harvest Street View images at each virtual camera location from Identify bent locations using stereo pair images A typical virtual camera configuration for a curved bridge Distance estimation for an object using a stereo pair image

78 Image to Model Determination of Deck Properties Determine deck type, top width of deck and year the structure was built from NBI Determine desk superelevation profile by combining geometry info. and speed limit data Estimate bottom width and height by utilizing fuzzy logic edge detection on harvested Street View images Estimate reinforcement detailing and corresponding structural properties

79 Image to Model Determination of Column Properties Determine the column type based on the number of detected column edges Sample column height and width at a number of levels Estimate rebar detailing and corresponding structural properties by interrogating a database of similar columns (and by utilizing Caltrans SDC)

80 Image to Model Determination of Column Properties Determine the column type based on the number of detected column edges Sample column height and width at a number of levels Estimate rebar detailing and corresponding structural properties by interrogating a database of similar columns (and by utilizing Caltrans SDC)

81 Image to Model Completion of model using crowdsourced data, metadata harvested from the databases Abutment types Column bearing types Shear key types and locations

82 Image to Model Completion of model using crowdsourced data, metadata harvested from the databases Abutment types Column bearing types Data to be refined by utilizing meta-data rules learned from Caltrans Standard Plans, Caltrans SDC via Deep Learning Shear key types and locations

83 Image to Model Completion of model using crowdsourced data, metadata harvested from the databases Abutment types Column bearing types Data to be refined by utilizing meta-data rules learned from Caltrans Standard Plans, Caltrans SDC via Deep Learning Shear key types and locations Complete model

84 Regional PBSA of Ordinary Caltrans Bridges

85 Location to Hazard Probabilistic Seismic Hazard Assessment (PSHA)

86 Location to Hazard Probabilistic Seismic Hazard Assessment (PSHA) A map of active faults around a Los Angeles site (Stewart, 2014)

87 Location to Hazard Probabilistic Seismic Hazard Assessment (PSHA) A map of active faults around a Los Angeles site (Stewart, 2014) Basic seismic hazard methodology (from Boore et al.)

88 Location to Hazard Probabilistic Seismic Hazard Assessment (PSHA) A map of active faults around a Los Angeles site (Stewart, 2014) Basic seismic hazard methodology (from Boore et al.)

89 Regional PBSA of Ordinary Caltrans Bridges

90 Analysis Models a brief overview

91 Building blocks of a bridge model

92 Building blocks of a bridge model Piles [Boulanger et al., 1999; Taciroglu et al., 2006; Khalili-Tehrani et al., 2014] Abutments [Stewart et al. 2007; Shamsabadi et al., 2010; Nojoumi et al., 2015] Shear keys [Mobasher et al., 2015; Omrani et al., 2015] In-span hinges [Trochalakis et al., 1997; Hube and Mosalam, 2008] Columns [Barry and Eberhard, 2008] Girders, deck (elastic)

93 Building blocks of a bridge model Piles [Boulanger et al., 1999; Taciroglu et al., 2006; Khalili-Tehrani et al., 2014] Abutments [Stewart et al. 2007; Shamsabadi et al., 2010; Nojoumi et al., 2015] Shear keys [Mobasher et al., 2015; Omrani et al., 2015] In-span hinges [Trochalakis et al., 1997; Hube and Mosalam, 2008] Columns [Barry and Eberhard, 2008] Girders, deck (elastic) Detailed descriptions of component and system modeling are provided in Omrani R, Mobasher B, Liang X, Gunay S, Mosalam K, Zareian F, Taciroglu E (2015). Guidelines for Nonlinear Seismic Analysis of Ordinary Bridges: Version 2.0, Caltrans Report No A0454, Sacramento CA.

94 Analysis

95 Analysis Monte Carlo (on cloud)

96 Analysis yields Monte Carlo (on cloud) IM-EDP Curve

97 Analysis yields Monte Carlo (on cloud) IM-EDP Curve Engineering Demand Parameter (EDP)

98 Analysis yields Monte Carlo (on cloud) IM-EDP Curve mean mean + std Engineering Demand Parameter (EDP)

99 Analysis yields

100 Analysis yields Fragility Curve

101 Analysis yields Probability of Collapse [or Probably of Exceedance of a predefined damage state for a particular component such as a shear key] Fragility Curve

102 Analysis yields Probability of Collapse [or Probably of Exceedance of a predefined damage state for a particular component such as a shear key] Fragility Curve

103 Loss Estimation an open problem for bridges

104 EDP or Performance State to Loss & Downtime

105 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties)

106 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings

107 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007)

108 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007) Packaged into FEMA Performance Assessment Calculation Tool (PACT)

109 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007) Packaged into FEMA Performance Assessment Calculation Tool (PACT) Provides fragilities/performance-functions for structural and non-structural components, and systems

110 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007) Packaged into FEMA Performance Assessment Calculation Tool (PACT) Provides fragilities/performance-functions for structural and non-structural components, and systems Generic FEMA P58 performance function Probability of Exceedance

111 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007) Packaged into FEMA Performance Assessment Calculation Tool (PACT) Provides fragilities/performance-functions for structural and non-structural components, and systems Generic FEMA P58 performance function Probability of Exceedance e.g., PoE of 2nd floor acceleration value of 0.4g

112 EDP or Performance State to Loss & Downtime Damage to a bridge leads to casualties and functional loss Direct losses (repair cost) and indirect losses (downtime and casualties) Extensive research had been carried out for buildings EDP to direct and indirect Losses (e.g., Porter, 2007; Mitrani-Reiser, 2007) Packaged into FEMA Performance Assessment Calculation Tool (PACT) Provides fragilities/performance-functions for structural and non-structural components, and systems Generic FEMA P58 performance function Probability of Exceedance e.g., PoE of 2nd floor acceleration value of 0.4g e.g., repair costs to MRI machine on floor 2 [other IQ examples are repair time, injuries requiring hospitalization, deaths, hours to operability, etc.

113 EDP or Performance State to Loss & Downtime

114 EDP or Performance State to Loss & Downtime Similar capabilities in loss estimation for bridges are lacking

115 EDP or Performance State to Loss & Downtime Similar capabilities in loss estimation for bridges are lacking Our long-term plans Try to replicate the FEMA-P58 methodology for bridges Develop apps (tools) for computing component fragilities (to enable rapid post-event assessment) Compile repair/downtime data and statistics (Caltrans) Devise methodologies for network impact and recovery analysis (UCLA Luskin Center)

116 A Validation Study San Bernardino I-10/I-215 Interchange Bridge Coronado Bridge, San Diego CA

117 Validation study San Bernardino I-10/I-215 Interchange Bridge

118 Validation study Selection of random points on the bridge by the user

119 Validation study Initial processing of selected points by program

120 Validation study Initial processing of selected points by program Calculation of bridge centerline curve *Using UCLA automated image-based structural model development program through utilization of

121 Validation study Initial processing of selected points by program *Using UCLA automated image-based structural model development program through utilization of Calculation of bridge centerline curve *Using UCLA automated image-based structural model development program through utilization of Elevation (m) Distance (km) Determination of ground elevations

122 Validation study Initial processing of selected points by program *Using UCLA automated image-based structural model development program through utilization of Road Elevation (m) Calculation of bridge centerline curve *Using UCLA automated image-based structural model development program through utilization of Elevation (m) Distance (km) Determination of road elevations *Using UCLA automated image-based structural model development program through utilization of Distance (km) Determination of ground elevations

123 Validation study Image processing to identify bent locations and developing the wireframe model

124 Validation study Image processing to identify bent locations and developing the wireframe model Identification of bent locations *Using UCLA automated image-based structural model development program via Image Analyzer Module

125 Validation study Image processing to identify bent locations and developing the wireframe model *Using UCLA automated image-based structural model development program via Wireframe Model Builder Module Identification of bent locations *Using UCLA automated image-based structural model development program via Image Analyzer Module Establishing of wireframe model

126 Validation study Image processing to identify in-span hinge locations Identification of in-span hinge locations *Using UCLA automated image-based structural model development program via Image Analyzer Module

127 Validation study Image processing to identify in-span hinge locations 1.5m Identification of in-span hinge locations *Using UCLA automated image-based structural model development program via Image Analyzer Module

128 Validation study Using of auxiliary data to determine superelevation profile* Identify centerline geometry in terms of constituent curves/spirals. Get bridge speed limit data through Google Roads API. Estimate curve superelevation at each sampling point. Determination of curve superelevation at each sampling point **Using UCLA automated image-based structural model development program via Image Analyzer Module Basic methodology to determine curve superelevation profile

129 Validation study Determination of bridge column dimensions Detection of column edges *Using UCLA automated image-based structural model development program via Fuzzy Logic Edge Detection Module

130 Validation study Determination of bridge column dimensions *Using UCLA automated image-based structural model development program via Pixel Counter Module Detection of column edges *Using UCLA automated image-based structural model development program via Fuzzy Logic Edge Detection Module Determination of column dimensions

131 Validation study Resulting model

132 Validation study Resulting model

133 Validation study harvested data vs. as-built: bridge deck elevation 1060 Deck Centerline Elevation (ft) Actual Measurements Harvested Data Distance Along Bridge (ft)

134 Validation study harvested data vs. as-built: column diameters Column Diameter (ft) Bent Number Cylindirical Part - Actual Cylindirical Part - Harvested Top Part - Actual Top Part - Harvested

135 Validation study harvested data vs. as-built: column heights Actual Measurements Harvested Data Column Height (ft) Bent Number

136 Validation study harvested data vs. as-built: modal periods T Image-Based (sec) T As-Built (sec) Mode Mode Mode Mode Mode Mode Mode Mode

137 Validation study harvested data vs. as-built: mode shapes Mode 1 Mode 2 Mode 3 Image- Based As-Built

138 Validation study harvested data vs. as-built: mode shapes Mode 1 Mode 2 Mode 3 Image- Based As-Built

139 Sample Applications Coronado Bridge, San Diego CA Wilshire Blvd/I-405N On-Ramp, Los Angeles, CA

140 Sample Application: San Diego Coronado Bridge

141 Sample Application: San Diego Coronado Bridge Selection of points along the bridge by the user

142 Sample Application: San Diego Coronado Bridge Initial processing of selected points by program *Using UCLA automated image-based structural model development code Road Elevation (m) Distance (km) Calculation of bridge centerline curve *Using UCLA automated image-based structural model development code Determination of road elevations *Using UCLA automated image-based structural model development code Determination of ground elevations

143 Sample Application: San Diego Coronado Bridge Using image processing to identify bent locations and developing the wireframe model

144 Sample Application: San Diego Coronado Bridge Using image processing to identify bent locations and developing the wireframe model *Using UCLA automated image-based structural model development program via Wireframe Model Builder Module Identification of bent locations *Using UCLA automated image-based structural model development program via Image Analyzer Module Establishing the preliminary wireframe model

145 Sample Application: San Diego Coronado Bridge Using image processing to identify bent locations and developing the wireframe model

146 Sample Application: San Diego Coronado Bridge Using image processing to identify bent locations and developing the wireframe model Final wireframe model

147 Sample Application: Coronado Bridge, San Diego

148 Sample Application: LA I10/I405N Interchange

149 Sample Application: LA Wilshire/I-405N On-Ramp

150 Testbed Route US-101/I-405 Interchange to Port of Los Angeles

151 Regional assessment

152 Regional assessment US-101/I-405 Interchange to Port of Los Angeles ~40 miles ~170 bridges

153 Regional assessment

154 Regional assessment

155 Regional assessment

156 What about Buildings?

157 Building models from image data

158 ShakeReady a user interface under development

159 Building inventories

160 Building inventories

161 Thank you!