Cost Effective Structural

cypriotcamelUrban and Civil

Nov 29, 2013 (3 years and 9 months ago)

75 views

Mecon

Limited

UCL May 2002

SIMoNet

Cost Effective Structural
Monitoring of North Sea Jackets

Using Acoustic Excitation to
Detect and Locate Damage in a
Multiply
-
Connected Structure


Dr Mark Harper


Mecon Limited, Cambridge

Mecon

Limited

SIMoNet

UCL May 2002

Current Approaches to Damage
Detection/Location


Modal response:
Simple in principle but limited
sensitivity.


ROV
-
intensive & diver
-
intensive methods. e.g.:

FMD

MPI

ultrasound (conventional contact method or
Lamb wave).

Generally expensive & time consuming.


Mecon

Limited

SIMoNet

UCL May 2002

Proposed Alternative: Continuous
Active Acoustic Monitoring


Used fixed transmitters and receivers to monitor
acoustic transfer functions


Identify time delays of changes and use to flag
up & locate changes in acoustic response of
structure


Test structure as frequently as required

(e.g. every 10sec)


Distinguish gradual from sudden changes


Reject periodic/temporary (e.g. tidal) changes

Mecon

Limited

SIMoNet

UCL May 2002

Can it work in a complex structure?


Model of upper half of
Claymore Jacket



Multiple acoustic wave types travelling at different velocities



Multiple acoustic paths & high levels of reverberation

Mecon

Limited

SIMoNet

UCL May 2002

Solution: Ignore all but direct
-
path
compression wave signals

Transmitter

Receiver

Damage
Point

Mecon

Limited

SIMoNet

UCL May 2002

Computer Simulation:

Forward Modelling

Model transit times of signals between all
nodes using modified time step method


Gain knowledge of propagation
characteristics


Compare with physical measurements


Use as basis for data inversion (identifying
& locating damage events)



Mecon

Limited

SIMoNet

UCL May 2002

Signal Characteristics

Examples: Transmitters 1,2 Receivers A,B Damage D



A

B

1

2

D

Mecon

Limited

SIMoNet

UCL May 2002

Characteristic model signals

100
200
300
400
0.02
0.04
0.06
0.08
0.1
Time
series
data
before
&
after
structural
damage
between
nodes
8
2
,
5
,
1
<
and
8
3
,
5
,
1
<
Transmit
from
node
8
1
,
5
,
1
<
Receive
at
node
8
4
,
5
,
1
<
Traveltimes
:
x
y
z
:
8
28.
,
25.4
,
12.7
<
Diagonals
:
8
37.8042
,
37.8042
,
35.921
<
Diff
After
Before
100
200
300
400
0.02
0.04
0.06
0.08
Time
series
data
before
&
after
structural
damage
between
nodes
8
2
,
5
,
1
<
and
8
3
,
5
,
1
<
Transmit
from
node
8
1
,
3
,
1
<
Receive
at
node
8
4
,
5
,
1
<
Traveltimes
:
x
y
z
:
8
28.
,
25.4
,
12.7
<
Diagonals
:
8
37.8042
,
37.8042
,
35.921
<
Diff
After
Before
100
200
300
400
0.025
0.05
0.075
0.1
0.125
0.15
Time
series
data
before
&
after
structural
damage
between
nodes
8
2
,
5
,
1
<
and
8
3
,
5
,
1
<
Transmit
from
node
8
1
,
3
,
1
<
Receive
at
node
8
4
,
4
,
1
<
Traveltimes
:
x
y
z
:
8
28.
,
25.4
,
12.7
<
Diagonals
:
8
37.8042
,
37.8042
,
35.921
<
Diff
After
Before
1
-
>A

2
-
>A

2
-
>B

Red line:

Pre damage signal

Green line

Post damage

Blue line:

Difference

Mecon

Limited

SIMoNet

UCL May 2002

Experiments

Plastic 1/100 scale model of Claymore Jacket

Damage Site 1

Main Leg

Damage Site 2

Diagonal Strut

Mecon

Limited

SIMoNet

UCL May 2002

Transmitters & Receivers


Transmitters: four piezoceramic cylinder
segments wrapped around each vertical leg


Receivers: piezoplastic wire helix wrapped
around each vertical leg


Sensitivity steered towards longitudinal waves

“Tomographic” geometry: transmitters at top,
receivers at bottom

“Reflection” geometry: transmitters and receivers
at top

Mecon

Limited

SIMoNet

UCL May 2002

Damage Location Algorithms:

Inverting Damage Event Arrival Times

1.
“Exact” inversion:


for each transmitter/receiver/event combination,
derive set of possible damage sites. Take intersection
of all of these sets to find damage site. Can use more
than one damage event per trace. CPU
-
time intensive
(millions of possible routes).

2.
“LMS” (least mean square) inversion:


use earliest damage event arrival only. Calculate mean
square difference between modelled and observed
arrival times for each node. Minimum value points to
most likely damage location. Fast.

Mecon

Limited

SIMoNet

UCL May 2002

Calibrating the numerical model

Pre
-
calibration: errors up to 30%

Post
-
calibration: errors typically 3%

Empirical travel time correction allows for unmodelled propagation effects and

systematic event time picking errors

-0.5
0.0
0.5
1.0
1.5
0
1
2
3
4
ms
ms
-0.5
0.0
0.5
1.0
1.5
0
1
2
3
4
ms
ms
Mecon

Limited

SIMoNet

UCL May 2002

Case 1: Damage to RH Rear Leg

LMS

Inversion of computer
-
model “tomographic”
data: Error Spheres

Error
spheres
for
each
node
position
on
structure
Transmitters
at
8
8
1
,
1
,
5
<
,
8
1
,
3
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
3
,
1
<
,
8
1
,
5
,
1
<
<
Receivers
at
8
8
4
,
1
,
5
<
,
8
4
,
2
,
5
<
,
8
4
,
4
,
5
<
,
8
4
,
5
,
5
<
,
8
4
,
1
,
1
<
,
8
4
,
2
,
1
<
,
8
4
,
4
,
1
<
,
8
4
,
5
,
1
<
<
Attenuation
delay
time
=
2.
ms
Attenuation
power
=
0.01
Minimum
LMS
error
of
0
occurs
at
node
8
8
2
,
5
,
1
<
<
Maximum
LMS
error
of
107.106
occurs
at
node
8
8
4
,
1
,
5
<
<
Damage Site 1

Main Leg

Mecon

Limited

SIMoNet

UCL May 2002

Case 1: Damage to RH Rear Leg

LMS

Inversion of scale
-
model “tomographic” data:
Error Spheres

Error
spheres
for
each
node
position
on
structure
Transmitters
at
8
8
1
,
1
,
1
<
,
8
1
,
1
,
5
<
,
8
1
,
3
,
1
<
,
8
1
,
3
,
5
<
,
8
1
,
5
,
1
<
,
8
1
,
5
,
5
<
<
Receivers
at
8
8
4
,
1
,
1
<
,
8
4
,
1
,
5
<
,
8
4
,
2
,
1
<
,
8
4
,
2
,
5
<
,
8
4
,
4
,
1
<
,
8
4
,
4
,
5
<
,
8
4
,
5
,
1
<
,
8
4
,
5
,
5
<
<
Attenuation
delay
time
=
1.
ms
Attenuation
power
=
4.
Minimum
LMS
error
of
3.44875
occurs
at
node
8
8
1
,
5
,
1
<
<
Maximum
LMS
error
of
91.532
occurs
at
node
8
8
4
,
1
,
1
<
<
Damage Site 1

Main Leg

Mecon

Limited

SIMoNet

UCL May 2002

Case 1: Damage to RH Rear Leg

LMS

Inversion of computer
-
model “reflection”
data: Error Spheres

Error
spheres
for
each
node
position
on
structure
Transmitters
at
8
8
1
,
1
,
5
<
,
8
1
,
3
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
3
,
1
<
,
8
1
,
5
,
1
<
<
Receivers
at
8
8
1
,
1
,
5
<
,
8
1
,
2
,
5
<
,
8
1
,
4
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
2
,
1
<
,
8
1
,
4
,
1
<
,
8
1
,
5
,
1
<
<
Minimum
LMS
error
of
0
occurs
at
node
8
8
2
,
5
,
1
<
<
Maximum
LMS
error
of
246.962
occurs
at
node
8
8
4
,
1
,
5
<
<
Damage Site 1

Main Leg

Mecon

Limited

SIMoNet

UCL May 2002

Case 2: Damage to LH Front Strut

LMS

Inversion of computer
-
model “tomographic”
data: Error Spheres

Error
spheres
for
each
node
position
on
structure
Transmitters
at
8
8
1
,
1
,
5
<
,
8
1
,
3
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
3
,
1
<
,
8
1
,
5
,
1
<
<
Receivers
at
8
8
4
,
1
,
5
<
,
8
4
,
2
,
5
<
,
8
4
,
4
,
5
<
,
8
4
,
5
,
5
<
,
8
4
,
1
,
1
<
,
8
4
,
2
,
1
<
,
8
4
,
4
,
1
<
,
8
4
,
5
,
1
<
<
Attenuation
delay
time
=
10.
ms
Attenuation
power
=
0.01
Minimum
LMS
error
of
0
occurs
at
node
8
8
2
,
1
,
5
<
<
Maximum
LMS
error
of
108.321
occurs
at
node
8
8
4
,
5
,
1
<
<
Damage Site 2

Diagonal Strut

Mecon

Limited

SIMoNet

UCL May 2002

Case 2: Damage to LH Front Strut

LMS

Inversion of scale
-
model “tomographic” data:
Error Spheres

Damage Site 2

Diagonal Strut

Mecon

Limited

SIMoNet

UCL May 2002

Case 2: Damage to LH Front Strut

LMS

Inversion of computer
-
model “reflection”
data: Error Spheres

Error
spheres
for
each
node
position
on
structure
Transmitters
at
8
8
1
,
1
,
5
<
,
8
1
,
3
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
3
,
1
<
,
8
1
,
5
,
1
<
<
Receivers
at
8
8
1
,
1
,
5
<
,
8
1
,
2
,
5
<
,
8
1
,
4
,
5
<
,
8
1
,
5
,
5
<
,
8
1
,
1
,
1
<
,
8
1
,
2
,
1
<
,
8
1
,
4
,
1
<
,
8
1
,
5
,
1
<
<
Attenuation
delay
time
=
10.
ms
Attenuation
power
=
0.01
Minimum
LMS
error
of
0
occurs
at
node
8
8
2
,
1
,
5
<
<
Maximum
LMS
error
of
136.994
occurs
at
node
8
8
4
,
5
,
2
<
<
Damage Site 2

Diagonal Strut

Mecon

Limited

SIMoNet

UCL May 2002

Conclusions


Both numerical simulations and scale model
experiments show active acoustic monitoring to
be feasible


“Reflection” geometry is:

-

potentially more accurate

-

more practical

-

most sensitive where damage most likely


Further scale model experiments planned


Next step: demonstration on a steel structure

Mecon

Limited

UCL May 2002

SIMoNet

Acknowledgement

Mecon Ltd gratefully
acknowledges the support of the
UK Health & Safety Executive