[ NDT-net Forum ]
[ Follow Ups ] [ Post Followup ] [ Help ]

Impact-Echo

Posted by: William Streett , E-mail: wbs3@cornell.edu, on May 12, 1999 at 18:05:22 :

I am writing in response to the article "Automated Impact-Echo: 2
and 3-D Imaging of Concrete Elements", by C. Colla, et al. in the
May 1999 issue of NDT.net. I am the co-author, with M. Sansalone,
of the book, "Impact-Echo: Nondestructive Evaluation of Concrete
and Masonry", Bullbrier Press, Ithaca, NY, 1997. (This book is
described on the web site http://www.impact-echo.com.)

The article by Colla, et al. is filled with so much misinformation
that one hardly knows where to begin. By ignoring a very large
body of published work on impact-echo, the authors have failed to
understand many fundamental aspects of the physics
of the method, and they have drawn conclusions that are unwarranted
and unsupported. Instead of taking the time to learn what is already
known about impact-echo, they have devoted time to the production of
of "pseudo 3-D re-elaborations" of impact-echo signals which they
have not correctly interpreted.

While the authors are certainly correct in pointing out that IE
can easily be misinterpreted, and that unskilled operators often make
mistakes, their description and analysis of their own IE tests
reveals a shallow and superficial understanding of the physics of
the method. They clearly do not understand how the choice of
equipment, including the transducer used to receive impact-echo
signals, the frequency range of the analog/digital converter, and the
sizes of the impactors used to produce stress waves, can affect the
results. They readily fit the description of unskilled and uninformed
operators of impact-echo.

On page 3 of their article they state that they used an A/D converter
with a frequency range of 10 Hz to 18 kHz. This suggests that the
instrument they are using employs an accelerometer as a transducer, but
this is not made clear. Regardless of the type of transducer used,
they have severely limited themselves by restricting
the frequency range to 18 kHz or less. The transducer developed
and used in the original instruments made by Sansalone and Carino
(reference 8 in the article), and used by other makers of impact-echo
instruments, is a broad-band transducer that responds to displacements
normal to the impact surface, and has an essentially flat response from
a few Hz to about 80 kHz.

By ignoring signals at frequencies above 18 kHz, they are unable
to identify signals from wave reflections at depths of about 10 cm
or less. In their "Results" section, for example, they state that
at the position of the pipe their signals indicated a greater
apparent thickness, but that the depth of the pipe was not indicated.
If the pipe is at a depth of about 9cm, as it appears to be, the
frequency of P-waves reflected from its top surface would be about
22 kHz. Naturally they cannot detect this part of the signal if
their upper frequency limit is 18 kHz. They could easily have avoided
this limitation through proper choice of equipment.

They refer to the importance of geometry in impact-echo testing, but
in designing their specimens #2 and #3 they have ignored one of the
most fundamental and important geometrical principles of impact-echo.
A fundamental rule of impact-echo, observed by Sansalone and others
more than 10 years ago, is that, for impact-echo purposes, structures for which
the lateral dimensions are less than about 5 times the thickness, are bounded
structures. Such structures respond in an entirely different manner
from plates. Specimens #2 and #3 are clearly bounded structures.
When a bounded structure is subjected to an elastic impact, cross-sectional
modes of vibration are excited, as a result of stress waves reflected
from nearby vertical boundaries. (If the lateral dimensions are greater
than about 5 times the thickness, the impact-echo response is
over before P-waves reflected from the side boundaries arrive at
the transducer, and only wave reflections through the thickness are
important.) The physics of the response of bounded structures was
worked out and published by Lin and Sansalone at the beginning
of this decade (Y. Lin and M. Sansalone, Journal of Acoustical
Society of America, Vol 91, pp. 895-893 and pp. 2474-2685, 1992.)
When cross-sectional modes of vibration are excited, additional
frequency peaks appear in the spectrum, but if the cross-sectional
dimensions are known, these frequencies can be predicted, and the
results can be interpreted accordingly. These cross-sectional modes
of vibration are almost certainly responsible for many of the "spots"
the authors report observing in their signals for specimens 2 and 3.
They correctly attribute these to geometry, but they fail to understand
that it is simply the effect of the geometry of their bounded
structure. If their specimen had been a large section of a
concrete highway, or bridge deck, with very large lateral
dimensions compared to thickness, the "spots" would have been
largely absent, and the determination of thickness and
depth would have been less ambiguous.

At no point in their article do the authors mention the importance
of the size of the impactor used in producing impact-generated stress
waves for impact-echo. The impact-echo method is based on the
physics of the impact of spheres on a flat surface, and the practical
method has been developed using small steel spheres (ball bearings),
typically mounted on the end of a spring rod, to produce the impact.
The physics of spherical impacts is discussed at some length in the
original work of Sansalone and Carino (reference 8 in the article)
and in the book cited at the beginning of these comments. The
diameter of the impacting sphere determines the distribution
of frequencies in the resulting stress waves, and the choice of this
diameter is a critical factor in successful impact-echo tests. The
sphere diameter determines the contact time -- the length of time
over which the sphere is in contact with the concrete surface -- and
this in turn determines the upper limit of useful frequencies in the
resulting stress waves. This information, together with many more
details of the physics of impact-echo, as well as practical aspects
of testing, are in the book cited above. A great deal of detailed
scientific work on the physics of impact-echo and on practical
aspects of impact-echo testing in the field, has been ignored in the
BAM-Berlin article.

At the very end of their article, Colla, et al. state, "Simulation of IE experiments
would be a very helpful tool for fully understanding the experimental
results." This is an astonishing statement, and reveals more than
anything else their ignorance of the field of impact-echo studies.
Beginning in about 1983, Sansalone adapted a finite-element program
developed at the Lawrence-Livermore Laboratories in the US, and
used it to simulate the propagation and reflection of elastic stress
waves in solids. In this method a continuum (concrete structure) is
divided into a finite number of discrete elements, and equations of motion for
the individual elements are combined to construct global equations describing
the dynamic behavior of the entire continuum. These equations are solved forward in time
using numerical methods. This work is described in detail in reference 8 in the article,
a US National Bureau of Standards Report that is essentially a reproduction of
Dr. Sansalone's PhD thesis. Indeed it was computer simulation of IE experiments
that guided virtually all of the subsequent development of practical applications
of the method, most of which the authors of this article have ignored.

There are other errors and misstatements in this article. Its publication
on NDT.net reveals the dangers in publishing papers that have not
been critically reviewed by others who are knowledgeable about the
topics covered.

Others interested in NDT methods for concrete and masonry would be well advised
to look elsewhere for additional informat



Follow Ups:


Post a Followup:

Name:
E-Mail:

Subject:

Comments:

Optional Link URL: Example: "http://www....html". Alternative write "http://www.ndt.net/wshop/attach/filename.htm" if you transfered a filename.htm via FTP
Link Title: is important! Information on my Home Page
Optional Image URL: http://www....gif


Attention! Please click the button only once. This may take a moment to process

[ NDT-net Forum ] [ Help ]