Infrasound Measurements of a Railroad Bridge
Dr. Mihan McKennaMs. Sarah McComas, Ms. Alanna Lester, and Dr. Paul Mlakar
U.S. Army Engineer Research and Development CenterGeotechnical and Structures [email protected]
Infrasound Technology Workshop 2008Bermuda November 3-7
Overview• Experiment
– Seismic, Infrasound, Acoustic and Meteorological measurements
– Load testing
• Modeling
• Data Analysis
• Future work
Motivations
• Two prior papers indicated that infrasound was generated by bridges at sufficient energy to be detected over background noise at certain times of the day/year.
• Traffic was not thought to be the source driver, perhaps natural sources such as wind excites the structure.
• Acoustic ducting was required for propagation of infrasound energy.
• Bridge may act as a dipole very close to source.
• Donn, W., N. Balachandran, and G. Kaschak. Atmospheric Infrasound Radiated by Bridges. J. Acoust. Soc. Am., Vol. 56, No. 5, Nov. 1974. 1367-1370.
• Kobayashi, Y. (1999) Infrasound Generated by a Highway Bridge. Butsuri-Tansa Vol. 52, No. 1. 54-60. (in Japanese)
Purpose:To determine the feasibility of remote
assessment of bridges using infrasound acoustics in combination with seismic, meteorological and audible acoustic methods.
Desired Payoff:• Field personnel can deploy small-aperture
infrasound arrays to ‘listen’ to a target structure and reliably analyze the situation without having to come into direct contact with the structure.
Research Goals and MotivationsResearch Goals and Motivations
Potential Results: Understand the physics of structure/atmosphere interactions resulting infrasound propagation. Foundation to create a catalog of bridge signatures to formulate algorithms for rapid remote assessment of infrastructure from bridges and other man-made structures.
Ft. Leonard Wood Deployment3 SIAM arrays
seismic, infrasound, acoustic and meteorological sensors.
2 standoff arrays
Array deployed at target bridge
30 meter
Airport Site2007-6-23
Actual sensor layout with scale
Fort Leonard Wood
Rolla
Test Area2007-June
Wastewater Range 19.9 kmAz 45 degrees
AirportAirportRange 26.867 kmRange 26.867 kmAz 39.4 degreesAz 39.4 degrees
The Infrasound Source Driver• Two 75 ton engines with eight flat cars of known weight
– Series of passes: eight, four, two, no cars with two engines, one engine, stopped and moving
• Controlled source with limited access during the experiment
•Generates the vibrational modes of the bridge used to discriminate against other background ‘noise’ including several other bridges in the area, both military rail and civilian interstate.
Meteorological Measurements– Three met stations deployed with one
per array consisting of temperature, pressure, wind speed, wind direction, dew point, humidity, soil moisture at two levels: 0.5 m and 2 m to estimate surface roughness.
– Five local environmental monitoring stations on post, recording temperature, wind speed, wind direction every 15 minutes at two heights: 3 m and 10 m.
– Collaboration with Hanscomb AFB for balloon radiosonde measurements. Total of five launches over the day of the test to ~ 30 km by state-of-the-art technology.
Meteorological Analysis– Only one inversion existed at the time of the train loading, at 0600
local time, before the test. There are no other ducting possibilities that day.
Propagation Modeling• InfraMAP modeling of the radiosonde data yielded only
one successful run, at 6AM local time. • Data analysis searching for the bridge signature will focus
on the time frame from 4 AM to 8 AM local time.
Integration of Source and Propagation Modeling
• Identified the optimal time for observing a possible signal from the target bridge: between 4 AM and 8 AM local time.
• What would the train signature look like?– Frequencies?– Continuous wave vs. discrete?– How does the source driver affect the signal?
13
Load TestingLoad Testing
• Bridge Description:– Type
• Pratt Truss (est. 1941)
– Material• Steel (built-up)
– Span• 7 Panels @ 23 ft.- 4 in.
– Height• 30 ft.
– Width• 15 ft.- 9 in.
– Skew• 65 ˚
14
Experimental MethodologyExperimental Methodology• Experimental load rating tests:
– Strain Gage (44 Used)• Main Structural Elements
– One Train Engine
Strain Gage
BDI
Auto ClickerBase Station
PC
Experimental MethodologyExperimental Methodology• Example: Strain Gages Results
Measurements
Str
ain
(
)
Experimental MethodologyExperimental Methodology• Example: Strain Gages Results
Measurements
Str
ain
(
)
COMPUTER MODELCOMPUTER MODEL
• Analytical Model (Frame):– Main Steel Structural Elements
– Built-up Sections
Stringer
Bottom Chord
Diagonal Chord
Diagonal
Floor Beam
Top Chord
• Mechanic of Materials• Stress
• =• Axial Force
• F = • Obtain Analytical Internal
Force
Summary of the Load Test and Analytical Model
Source Modeling
• COMSOL Multiphysics Structural Mechanics Module• Key components
– Simplified source to limit computational cost in large model– Accurate represents sound emitted from bridge
• Technical difficulties– Bridge models to determine natural frequencies typically
constructed using beam/truss elements– Beam/truss elements appear as point sources in acoustic
analyses– Geometry of beam important for acoustic response– Natural frequencies of bridge do not provide obvious
simplification of bridge structure – no single area dominates acoustics (e.g., bridge deck)
Ft. Leonard Wood Bridge
• Pratt Truss Bridge• Struts included by specifying equal z-displacements at top
of vertical member pairs
z
z1
z2 21 zz
x
Natural Frequencies - Overview• Bridge shows 230 modes between 2 and 20 Hz• Three General Categories of Modes
– Category 1: Relatively large deformation of many components (10%)
– Category 2: Relatively large deformation of few components (33%)
– Category 3: Relatively small deformation of components (57%)
• First two categories should dominate acoustical energy
Natural Frequency: CategoriesRelatively large
deformation of a few components
Relatively large deformation of many
components
Relatively small deformation
11.9 Hz
13.6 Hz 10.0 Hz
All results plotted w/ same deformation
scaling factor
Natural Frequencies – Observation 2 hz
• Modes show deformation in z direction, stringer: stays in plane
• No modes show significant deformation in y direction• Bridge design requires large stiffness to resist
deformation in y direction (designed to prevent cantilever bending)
xz
Methodology
• Rank bridge components based on source strength– Cross sectional area (CSA) perpendicular to direction of motion– Relative acceleration identified from natural frequency analysis
• Develop acoustic model of critical components of bridge using shell elements of CSA– Develop detailed model using simple shape of beam CSA – Apply deformation mode from natural frequency analysis to each
component
• Use solution on outer boundary of acoustic model to drive infrasound solution over large domain
Effect of CSA on Acoustics
10 cm
4.7 m
Deck ModelBeam Model
Models excited using same
accelerations150 m from source,
normal above (y)
Geometry Effect
CSA Source Modelcenter to center spacing, real measurements
• Representation of stringers (plan view)
• Apply source acceleration in y direction (n = 1→5)
10 cm
10 cm
10 cm
10 cm
76cm
x
z
Comparison with Continuous Model
• Do small gaps between stringers affect results even at 1 Hz (l=343 m, gap=0.76 m)?
• Small gaps (relative to l) affect acoustic response YES
Radiation Pattern4 Beam Model
3 4
Summary
• Ft. Leonard Wood bridge shows complex frequency response
• Cross sectional area of beams has strong effect on acoustics
• Small gaps relative to wavelength have effect on acoustics
• Simple representation of bridge deck shows strong directionality
SIAM Data, raw infrasound
IML Airport
IML Airport
IML Airport
IML Bridge
IML Bridge
IML Bridge
IML Bridge
Interference infrasound generated by the train
What frequencies does the train generate? – Bridge Array
Bridge Array – 3020 SW of Bridge, closest to train at this time
Bridge Array – 3019 – NE of Bridge, further from train
50hz 40h
z
From 18:34:00 to 18:34:30 UTC
[13:34:00 to 13:34:30 Local]
What frequencies does the train generate? – Spectrogram of WTF and Airport Arrays for 18:34:00
to 18:34:30 UTC, [13:34:00 to 13:34:30 Local]
3012 – Airport Array 3015 – WTF Array
40hz
WTF has frequencies up to and including 40hz
Airport has low frequency (up to 6hz) and then high frequency (80hz)
80hz
10hz
Bridge signal from WTF array
Difficult to find arrivals in signal
2hz signal in time series
FK analysis results correspond with bridge azimuth
• Includes frequencies of interest (2, 10 and 13hz) - extremely low amplitude
• Time series does not appear to have high activity
• Additional higher frequencies (42 and 56hz) present and other additional transients (>50hz)
• Higher frequencies (40 – 50hz) are persistent through the two hour time period
Greater dynamic of frequencies present at airport
Quiet times are clearer than WTF array (30hz signal visible)
Loud times have greater frequency range
1005 UTC
(0505 AM Local)
1052 UTC
Bridge signature is clearer at the airport array despite the changing dynamics at the airport
Airport array frequencies
FK analysis indicates correct azimuth and apparent velocity
Bridge signature from Airport Array
• Station 3010 is not usable due to faunal mastication (rabbit) during data acquisition
• Frequencies of interest present – with higher amplitudes than at WTF array
• Time series shows that the airport array is more active than WTF array
Future Modeling and Data Synthesis
• The finite element model of the bridge created during the load testing will be uploaded to a multi-physics finite element package.
• The bridge will be coupled into the atmosphere and vibrated at the frequencies observed during the test.
• A representative source ‘package’ will be developed for use in infrasound propagation modeling software.
Questions?