Background

Notes from the 2017 BBD workshop held at the Kiewet Training Center in Omaha NE.  NDS attended the workshop as part of our collaboration with the BBD community for the NBI pilot and our collaboration with MBDH.

https://bridgingbigdata.github.io/pages/bbd2017.html

Summary

The BBD workshop was structured as as set of presentations and breakout groups focused on specific themes. Themes included data management, decision support systems, socio-technological impact, and next-generation health monitoring. NDS is collaborating with Robin Gandhi on the NBI pilot and participated in the data management breakouts.

The goals of the workshop included building a community of practice, facilitating data discovery and controlled sharing, and identifying opportunities for applications of big-data technology in bridge health monitoring. The BBD community seems to span commercial, public, and academic stakeholders from areas including computer/information science, civil/structural engineering.

For data management, they are working with HubZero, which doesn't currently support in-browser analysis or notebook sharing. Security and access control will be essential, since bridge data is often owned by asset owners, not the public.

Today, the FHWA NBI data is one of the more important publicly available resources.  The NBI pilot will demonstrate common analyses using this data.  However, UAV and sensor-based inspection will see increasing adoption and they will require both the ability to manage these data pipelines and perform analysis.  On the civic side, nearly all states use the AASHTO BRM (Pontis) system.  This collects local asset and inspection information and is rolled up into the public NBI data.

The data management theme focused on challenges in managing data about rural bridges (part of the spoke focus).  Key challenges identified by participants included training/education/expertise development (even in the limited context of the AASHTO system); questions of funding/resources (how to pay for collecting information); institutional barriers to adopting new technologies/workflows; questions of data quality and interoperability. The discussions were dominated by civic IT staff working with the AASTHO BRM system.  However, the facilitator did hope to move the discussion to questions about next-generation monitoring – how to instrument old bridges and what to do with the data; public data collection etc.


Detailed notes

Notes

~30 attendees

Welcome/intro

  • Organizers
    • Dan Linzell (Civil Engineering Department Chair, UNL)
    • Robin Gandhi (Information Science, UNO)
  • Hesham Ali (Dean, College of Information Science and Technology)
    • Emphasis on safety in the context of recent natural disasters
    • Infrastructure safety issues can be addressed
    • Also performance, optimization, cost.
    • Very interested in collaboration
  • Mark?
    • UNO emphasis on big data research
    • Moving discussion from data to knowledge and decisions
  • Melissa Cragin (MBDH/NCSA)
    • Overview of big data hubs
    • Goal: drive innovation in big data space
      • Catalyzing public/private partnerships; building out national data infrastructure
    • Mention of hackathon, should collect what they did

BD Spokes project updates/results

  • Robin/Dan
  • Goals:
    • Build a community of practice
    • Identify opportunities for big data technology for bridge monitoring
    • Data discovery and controlled sharing. Sharing is critical.
  • Collected some data
    • Surveys to state DOTs and county/city engineers to understand how data is collected related to bridge health
    • Solicited data
    • Many concerns about security
  • Next generation health monitoring
  • Working session results – BBD 2016
    • What are the challenges
      • Inconsistent format
      • Understanding value of data
      • Interpretation/misinterpretation
      • Security
      • Amount of data 
    • Benefits of sharing data
      • Decision making
      • Cost savings/ROI
      • Safety
      • Sharing new ideas
      • Transparency to end user (public): comparison to quantified agriculture
    • Decision support
      • Summary: managing larger and larger amounts of information
      • Two breakout sessions (issues/concerns; where can we go)
    • Standardization for bridge health data
      • Existing standards
      • Improved standards
      • User focus
      • Agreement
    • Socio technical impact
      • Summary: How to change current practices and communities
  • SMARTI: Smart big data pipeline for aging rural bridge transportation infrastructure
    • Goals
      • Aging rural bridges
      • Mine existing and new datasets using next-gen health monitoring techniques
      • Reference implementation of a big-data pipeline
      • Build useful data products
    • Reference cases
      • Identify representative structures in rural Nebraska to use as focal points for generating data and utilizing data
      • Demonstrate effective data management, monitoring, impact
      • Possibly mine existing datasets (partners: Union Pacific, Nebraska DOT)
    • Production of data products
      • Invite people to create new data products
    • Areas:
      • Next-gen monitoring
        • Low cost systems
        • UAV
        • Fusion techniques to create/discover new datasets
      • Data management
      • Decision making
        • Prediction
        • Modeling
        • Visualization
      • Social-technical impact
        • institutional challenges


Breakout group: Data management

  • Prompt:  Challenges with bridge health data management in rural areas; asset owners in rural areas
    • Each participant asked to think about challenges; round the room
    • High-level areas:
      • Training/Education/Expertise development
      • Cost/funding/resources - sustainability
      • Workflow/use cases
      • Data - quality, interoperability, etc.
      • Institutional barriers


Keynote: Framework for risk-based inspection (RBI) practices

  • Glenn Washer, U Missouri, Civil Engineering
    • Thermography, radiography; visual inspection; 
  • NCHRP 12-82; FHWA national bridge inspection standards
    • Goal; improve safety and relaibility of bridges
    • Focus inspection where needed
    • Optimize use of resources
  • Inspection of structure is a big field
  • Infrastructure is behind other fields (e.g., nuclear power, pipelines, etc)
  • NBIS standards (1971) – 2 year cycle; uniform guildelines
    • Different needs for different regions
    • Need knowledge from regions and people who own the bridges
  • Described process for NBI
    • Estimate probability of failure off of a variety of events.
    • Time in condition rating
  • What was the data, methods etc?
    • Structured questionnaires working with local experts (qualitative)
  • Role of data
    • Deterioration modeling
    • Health model/remote modeling
  • Q. What did you use to develop the models?
  • Q. What did you do with the data?
  • Questions
    • How many states have gone to 48 month inspection? ~15 or so
    • Question about new inspection practices
    • Discussion of international scan – qualify expertise and level of inspector; cost of inspection;
  • Q. Data/analysis workflow
    • Used NBI data – but other data could be used
    • Cleaned/processed to fit assumptions
    • Statistical analysis via Excel, etc. – but other methods/models/software could be used (if you could format the data)
    • Sharing is public because it's published as a report
    • FHWA produced some new data

Demonstrations:

  • Crack detection using car mounted and UAV sensors
  • Point cloud analysis
  • MongoDB piplines for NBI data
  • Note: They are using  datacenterhub.org at Purdue (HubZero)


LaViolette: Bridge Data, BRIM, and SHM: Real World Experiences

  • Pennsylvania Bridges
    • Many old bridges
    • Low hanging fruit: find and replace old bridges
    • P3/design build program to replace faster
    • 558 bridges
  • Data
    • Bridge type
    • MOAS: Highly sophisticated spreadsheet ("Mother of All Spreadsheets")
      • Preliminary design of all bridges quickly
      • Calculated estimated material quantities for bidding job used by contractors
    • Data management from contracting standpoint
  • Process
    • Design team (25+ squads, 365 FTE at peak)
    • Many moving parts; Bluebeam (editing/reviewing of drawings, multiple timezones)
    • RBRP Web Application
      • Keep track of assignments; dashboards; tools
    • Project management/project planning
  • Q. Does data feedback into the project management?
  • Part II. BridgeAdvise - prioritization using risk factors
    • Software for prioritizing bridge replacement
    • Probability of failure and consequence of failure
    • Data:
      • Patented algorithm
      • NBI or PONTIS data
    • Web-based application provides visualization
    • Tale of two bridges:  same POF, different COF
  • Which bridges are best for accellerated bridge construction (ABC)?
    • ABC Rating score & flow chart
      • Only data available via NBI (average annual daily traffic, out of distance travel, daily road user costs, economy of scale)
    • AHP
  • Presentation of New Tappan Zee Bridge
  • Model and data aggregation
    • Autodesk/Berkeley – but who knows in the future
    • Try to tie everything to a 3d model of the bridge
  • Intelligent transportation systems
    • Connect weather to speed signs
    • All electronic toll collection
  • Conclusions
    • Use of brigdge data is growing part of the profession
    • Data helps bridge owners make informed decisions, but it needs to be practical
    • Practical uses for developed systems
  • Q. Size of sensor data?
    • Very large. 25 sensors collecting constantly.
  • Q. Process of sharing data
    • Throughway is protective of anything related to that bridge
    • Open to the idea of sharing the data, but need to understand how they can use it
  • Q. Need to make informed decisions
    • Which bridges need work first? Every state has a different approach. If we could come up with a uniform way to help owners make decisions would be a great place to start.
  • Q. Risk-based prioritization for replacement, but what about preserving high-value assets
    • Four data points for each bridge, could use more data if not more noise

Breakout questions:

  • What data is important for sharing?


Afternoon:

  • Challanges: What are effective solutions for data management for rural bridges for asset owners?
  • Theme:
    • Hosting services  (for counties)
    • Reference implementation
    • Google maps for bridges
  • Q. What is the data?
    • Inspection records; sensor data; design data; planning data; bid data; 
    • Some private; some public
    • Daily traffic
  • Ideas:
    • Make sure they have computers, email, internet (they loan out laptops)
    • BRM inspectors might have access at different levels (national database)
      • If you're collecting outside of what BRM is asking for
      • Often manage many assets (traffic lights, etc)
    • Implement AASHTO BRM and BRR http://aashtowarebridge.com/
      • They see a webpage. They can select bridges that they're authorized to see. They can export to Excel/KML
      • Let them configure what they would want for work candidates
      • Keep results associated with structure number
      • Element based (inspection)
      • NBI (general condition inspection)
      • Coarse at administrative level
        • No defect/location based metric
      • BRR is load-rating tool
        • Much more refined
    • Excel tool to support county bridge decisions with data refreshed by BRM 
    • What about next-gen sensing
      • Citizen data, sensors, UAV, etc.
    • Who owns the data – how they provide access?
    • The data is available through FOIA
      • Detailed work programs and photos are on FHWA system
      • Not private, not < 20 ft.
    • If we have low-cost sensors for off-sensor bridges, are there existing systems?
      • Continuous sensing – no one wants year-round data, but they do want threshold instrumentation
    • You're submitting NBI data, not BRM
      • Next year data format will change
    • Working with local stakeholders to figure out what their needs are
    • Funds for resources to collect the data
    • Requirement to collect additional information (elemental data)
    • Model
      • Formal definitions of elements (different limit states) codified in specifications or standards independent of the software
      • This is what the state bridge community uses
      • AASHTO has a branch doing software devevelopment for bridges and other assets (10x size of organization)
      • They hold copyright/licenses for three different products: BRM, BRD, BRR
      • Software task force from states
    • BRM user group
    • Where would sensing data go?
      • BRM would be a good place for it
      • Show that large farm vehicles work
    • Make form for county superintendents to share tips/tricks that work for them
  • Thoughts
    • Data management is research data management and models used to inform public infrastructure
    • Prototype data and computational models
    • Should inform BRM user group decision making for new features
    • Recognized practical importance of BRM systems

UAV/UAS for infrastructure monitoring session

  • Support by Nebraska system sceince planning grant: autonomous structuraal health monitoring of critical infrastructure using intelligent multi-UAS
  • Alexis: Robotic inspection for infrastructure facilities and nuclear sites
    • What can be achieved when multiple modalities can be fused
    • How can we select what the robot observes (small endurance)
    • Multimodal localization
      • Fused: stereo camera, inertial measrument, 3d depth sensing (synchronized)
      • Needs to be fast (must make decisions based on fused data)
    • Scanning for things that are more important (new information)
    • Robot estimate radiation sources
  • Moreu: Using unmanned aerial systems (UAS) with lasers to assess structure performance
    • 100,000 railroad bridges in the country
    • Priorities of owners: cost-effective maintenance of bridges
    • Part of SMARTI
    • What about pedestrian bridges
    • Research question: how to detect bridge movement (to slow the train if needed to cross the bridge), measure displacement
      • Wireless sensor for reference-free displacement measurment – requires installation (100,000 bridges...)
      • Laser (cost and size)
    • UAV – may be useful later (UAV + laser)
      • Multiple drones/lasers considered
    • Remote tapping 
    • Reference (2025 Vision from ASCE)
      • real-time access to living databases, sensors, diagnostic tools to make informed decisions.

Summary:

  • Decision support breakout
    • What are the challenges? What's the right architecture to aggregate this data? etc
    • Parameters, proactive, performance: 
    • Distinction between "researchers" and "owners"
    • Before there's a hole in the deck of the bridge, know that there's a hole


Thoughts for Workbench:

  • HubZero doesn't allow in-browser analysis and notebook sharing. Can this feature (Workbench) be added to HubZero?
  • Security, permissions, etc. will be essential to this community:
    • Ability to combine commercial datasets with public datasets
    • The bridges are owned by asset owners, not always public
    • NBI data is public, but even local DOT data isn't
    • NBI data is core to the community, along with AASHTO BRM (Pontis) system data
  • Similar request from TERRA-REF team – clear method for sharing notebooks/applications between users
  • UNO is doing more sophisticated analysis with Mongo pipelines, but also interested in Spark.
  • BBD project has requirements tension between existing systems/formats (e.g., BRM/NBI), emerging technologies (UAV, sensor-based inspection) and research needs.


  • No labels