acoustic emission monitoring - university of delaware
TRANSCRIPT
The World Federation of NDE Centers
Acoustic Emission monitoring
Thomas SchumacherUniversity of Delaware
E-mail: [email protected]
Short course on NDE for the infrastructureBurlington Vermont, July 16th and 17th, 2011
� Overview of fundamental basis
� Overview of technology
� Review of latest developments
� Strengths of method
� Limitations of method
� NDE application: case studies
� Summary and conclusions
Overview of presentation
2
� Acoustic Emission (AE) is the term used for transient elastic waves generated by the release of energy within a material or by a process (EN, 2000).
� Irreversible process� Source time, location, and mechanism
unknown� Passive technique� Sensing via surface-mounted piezo-
electric transducers� Similarity to earthquakes, i.e. nano-
seismic activity� Frequency range of AE in concrete:
~10 to 500 kHz
Overview of fundamental basis
3
Medium
Sensor
to DAQSource
External load
� Primary sources� Micro-cracking (distributed)� Macro-cracking (localized)� Compression failure (crushing)� Yielding and fracture� De-bonding between materials
� Secondary sources� Sliding/friction between interfaces
� Artificial sources� Calibration sources (pencil lead break, ball drop, pulse)
� Noise� From bearings, supports� Background: ambient traffic, vibrations� From electrical circuit, cell phones
Overview of fundamental basis (cont.)
4
Schumacher, 2008
Angerinos et al., 1999
� Elastic waves in finite media (non-dispersive)
� Reflected/diffracted waves� Guided waves in plate-like members (dispersive)
� Plate waves� Lamb waves
� Wave attenuation� Geometrical� Scattering� Internal friction
Overview of fundamental basis (cont.)
5
Compression wave (fastest) Shear wave Surface wave (slowest)
Frequency, f [kHz]
Nor
mal
ized
am
plitu
de [-
]
0 100 200 300 400 5000.0
0.2
0.4
0.6
0.8
1.076 mm (3 in.)
Frequency, f [kHz]0 100 200 300 400 500
152 mm (6 in.)
Frequency, f [kHz]0 100 200 300 400 500
305 mm (12 in.)
Frequency, f [kHz]0 100 200 300 400 500
1143 mm (45 in.)
Increasing travel distance
Adapted from Wood: http://www.geo.mtu.edu/
Schumacher, 2008
� Measurement process
Overview of technology
6
Source: Ch. Grosse, TUM
b-Value
� Model of the measurement process
Source signal, S(t)�
Stress wave�
Propagation�
Surface motion � voltage�
Amplification�
Filtering Response function�
Digitization/storage on PC�
Response signal, R(t)
Overview of technology (cont.)
7
Pre-amplifier, tfR(ω)
Source, S(ω)
Stress wave front, p-wave
Sensor , tfS(ω)
Data acquisition system , tfR(ω)
Medium, tfG(ω)
( ) ( ) ( ) ( ) ( )G S RR S tf tf tfω ω ω ω ω= ⋅ ⋅ ⋅
( ) ( ) ( ) ( ) ( )G S RR t S t tf t tf t tf t= ∗ ∗ ∗⇕
Adapted from Schumacher, 2008
� Sensors� Piezo-electric (PZT) devices � Voltage output proportional to surface motion� Resonant vs. broadband� Coupling
� Pre-amplifiers� Amplify small sensor output
� Transient recorder� 14 to 18-bit dynamic range typical� Recording rates ≤ 40 MHz (practical ≤ 10 MHz)� Analog filters� Parameter extraction� Full waveform storage� Independent recording using trigger criteria
Overview of technology (cont.)
8Source: Vallen Systeme GmbH
Schumacher, 2008
1)Fowler et al., 1989, 2)Ohtsu et al., 2002, 3)Gutenberg & Richter, 1949,4)Grosse, 1996, 5)Geiger, 1910, 6)Aki & Richards, 1980
� Overview methods of analysis
Overview of technology (cont.)
9
Stored AE signals, R(t)
AE event forming
Qualitative Quantitative
Source parameters5):- Location
- Time
AE parameters- Hit rates/energy/…
Waveform analysis:- Comparisons4)
Moment Tensor Inversion6)
- Historic-severity1)
- Load-Calm ratio2)
- b-Value analysis3)
� Qualitative� Statistical analysis of AE parameters� Does not relate observations with physical parameters (source mechanisms)� Can be performed with as few as 1 sensor� Readily available and implemented in commercial AE systems� Relative measure, only comparable if exact same conditions� Depend on selected acquisition and threshold criteria
� Developed methods� Load-Calm ratio (Ohtsu, 2002)� Historic-Severity index (Fowler, 1989)� b-Value analysis (Gutenberg &
Richter, 1952)
Overview of technology (cont.)
10
Source: ASTM E602 (1982)
� Qualitative (cont.)� Kaiser Effect (Kaiser, 1950): In most metals, AE are not observed
during the reloading of a material until the stress exceeds its previous high value.
� Felicity Ratio (Fowler, 1986): Break down of Kaiser Effect due to material instability where AE start to occur before its previous high value is reached.
Overview of technology (cont.)
11
Koeppel, 2002
Overview of technology (cont.)
12
� Qualitative (cont.)� NDIS-2421 (Ohtsu, 2002)
� Historic-Severity Index (Fowler, 1989)
� Problem: selection of triggerinfluences results!
Golaski et al., 2002
Ohtsu, 2002
Schumacher, 2008
Overview of technology (cont.)
13
� Qualitative (cont.)� b-Value analysis (Gutenberg & Richter, 1949)
� Waveform correlation (Grosse, 1996)
2 2.5 3 3.5 4 4.5
0
0.5
1
1.5
2
AE Magnitude [AdB/20]
log(
Cum
ulat
ive
AE
Hits
) [-
]
Frequency distribution of hit amplitudes
Estimated b-value (slope of this line)± one standard deviation of data
Data mean value
50 hits
Amax
Grosse, 1996
Magnitude-squared coherence
� Quantitative� Relates observations with physical parameters (source mechanisms)� Requires a network of sensors (≥ 6 for moment tensor inversion)� Requires data with high signal-to-noise ratio� Difficult to apply (complicated procedures, still in research stage)
� Source Locations� Arrival time difference method (Geiger, 1910)� ≥ 4 sensors� Accuracy from outside sources low
Overview of technology (cont.)
14
81
23
4
5
6
7
p-wave front
1st hit sensor
AE source
600 650 700 750 800 850 900 950 1000-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
Sample # [-]
Sig
nal a
mpl
itude
[m
V]
/ A
IC f
unct
ion
valu
e [-
]
Original Signal
Filtered Signal
AIC Function (on Filtered SignalFloating Threshold Picker
AIC Picker
Schumacher, 2008
� Moment Tensor Inversion (MTI) (Aki & Richards, 1980)� Source mechanism� Requires ≥ 6 sensors� Pre-requisite: accurate
locations (to computeGreen’s functions)
� Radiation pattern inferredthrough surface observations
� Problematic for crackedspecimens (high non-homogeneity)
� Knowledge of responsecharacteristics of systemcomponents required
� Need to use high-fidelity sensors
Overview of technology (cont.)
15
Grosse et al., 2003
Grosse et al., 2003Sansalone, 1997
� Moment Tensor Inversion (MTI) (cont.)
Overview of technology (cont.)
16
Grosse et al., 2001
Shigeishi et al., 2003
� Development of high-fidelity sensors (e.g. Glaser-NIST)
� Sensitive, extreme broad-band, absolutely calibrated
� Wireless sensor networks� Array techniques� High accuracy outside sources
Review of latest developments
17
Grosse et al., 2004McLaskey et al., 2007
Source: KRN Services
� More robust hybrid Moment Tensor Inversion (Linzer, 2001)
� Combines absolute and relative MTI(relative: No need to compute Green’s functions)
Review of latest developments (cont.)
18
Linzer, 2001
Linzer, 2001
� Probability based source location algorithms (Schumacher, 2010)
� Use of seismology based methods for quantitative analyses� New location methods, MTI, moment magnitude, tomography
Review of latest developments (cont.)
19
Schumacher, in review
Strengths of method
Advantages:� Applied during testing/loading� No disturbance during application� Real-time feedback� Detection AND characterization of
internal fracture processes as theyoccur
� Covers volume (distributed sensing)
Useful for:� Monitor progression of existing damage (e.g. crack propagation)� Real-time detection of occurring overloads (alarm system)� Continuous (long-term) monitoring of critical components� Verification of retrofits and repairs (before/after)� Complimentary for in-service load testing (Acoustic Emission Testing)
20
Katsaga et al., 2007
Limitations of method
21
� Very few standards available for infrastructure (NDIS-2421, RILEM)� Large variability in structures (type, geometry, material properties)� Complexity of structures and components � Changing boundary conditions (e.g. cracking or sensor coupling)� Tests not truly reproducible due to nature of AE
� Cannot tell current state such as existing cracks, only change in state
� Background noise can be significant = low signal-to-noise ratio� High variability of signal strengths
� Quantitative analyses often difficult to apply in real-world situations� No long-term monitoring experience with this method
NDE application 1: pressure vessels
� Well established, confidence high� Large pool of samples – baseline data available� Well-defined problem (geometry, material properties)� Loading protocol established� Loading known –
applied pressurecan be easilycontrolled
� Analysis method:historic-severityindex (Fowler, 1989)
22
Catty, 2010
NDE application 2: laboratory RC beam
� Large-scale experiment on RC beam using quantitative analyses(Katsaga et al. 2008)
� Source parameters� Moment Tensor analysis� Insight into development
of fracture during loading� More shear type sources
in the later loading stages
23
Katsaga et al., 2008
NDE application 3: wire breaks on bridge
� Continuous monitoring of post-tensioned bridge (Fricker & Vogel, 2006)
� Monitoring for steel wire breaks� Verified by induced breaks
(after bridge decommissioned)
� Ideal application: sources ofinterest1) high energy comparedto other sources2) and noise
� Example of alarm system
24
1) 2)
Fricker & Vogel, 2006
Fricker & Vogel, 2006
� RC deck girder bridge in Cottage Grove, OR (Schumacher, in preparation)
NDE application 4: in-service load test
25
Crack displ.
AE sensors
Strain gage
� RC deck girder bridge in Cottage Grove, OR (cont.)
NDE application 4: cont.
26
Qualitative Quantitative
NDE application 5: retrofit of steel bridge� Noisy bearing of swing bridge in Reedsport, Oregon
� AE activity during operation before/after replacement of bearing
27
Summary and conclusions� Passive method for monitoring of fracture processes� Applied during testing/normal operation – real-time feedback
� Useful for monitoring and as alarm system:� Prestressed concrete beams� Crack progression monitoring� Location of mechanical noise
during operation� Fracture monitoring during
experiments
� Promote use of principlesfrom seismology forquantitative AE
28
Source: ITI, Northwestern University (website)