Cambridge Ultrasonics (CU) and the Institut
fur Massivbau (IfMD) at the Technische Hochschule Darmstadt, Germany,
have been given a strong endorsement to proceed with a development
project 
being
jointly promoted in the field of ultrasonic inspection of large
concrete structures. A meeting was held in Darmstadt in May for
potential partners in the project at which there was strong support
for both the technical approach proposed and the project structure.
Three organisations had registered to join the project before
the meeting, one organisation agreed to join the project at the
meeting and since then one university has joined as a full partner,
making five partners in addition to IfMD and CU. Several other
organisations are considering participating including the German
Concrete Association. An application will be made for the project
to get Eureka status; it is currently described in the Eureka
web site (http://www:eureka.be).
The project is known as the second generation project because it will build upon the results of two previous projects in which CU has taken a leading role. The best aspects of each project will be developed to a stage where prototypes are ready for commercialisation as instruments and as services in the hands of inspection service providers. The instruments resulting from the project should find application in a wide range of structures including: offshore oil production, nuclear power stations, dams, bridges, roads, telegraph poles, cat cracker linings and liquid petroleum gas confinement slabs.
IfMD co-ordinated one of the first generation projects sponsored by the German Concrete Association. Starting in 1993, the project created a novel ultrasonic pulse-echo system for concrete with a breakthrough in performance. Targets of 20 mm diameter (reinforcement bars) could be detected at a range of 0.5 m with a success-rate of typically 90% and an axial accuracy better than +/-5%; larger targets (stressing tendons, back walls) could be detected at even greater range. The greatest range achieved so far is approximately 1.8 m. The system has several novel features including: transducer arrays and computer signal processing based upon a personal computer. The present laboratory prototype has worked well in a wide range of experiments, all from a single test surface: detecting stressing tendons, reinforcement bars, honeycombing, measuring beam thickness and screed adhesion quality. Unfortunately, the prototype is not particularly suited to site testing because it is not easily transported. One of the objectives in the second generation project is to build a new version that is suitable for use on-site. More evaluation experiments will then be possible and a second objective is to assess performance on a wide range of structures owned by partners in the project. Development of a site-worthy instrument, although involving considerable work, is seen as relatively low-risk - mother nature is not being challenged. This is an example of why the project offers a good chance of achieving success.
CU co-ordinated the other first generation project, which was supported by the European Commission under its Brite/EuRam programme (BE-5029-92). The objectives were to improve the method of resonance spectroscopy and to evaluate it on a range of industrial materials. The project was successful in developing new equipment and an artificial neural network was found to interpret the quality of test samples (finding cracks, dimensional errors and material property deficiencies). Whenever the neural network was used the success-rates in correctly classifying the quality as pass or fail were in the range 87% to 100%. On metallic and pre-cast concrete samples the technique worked very well but when interpreting data without a neural network it was impossible to assess the quality. One conclusion resulting from the project is that resonance spectroscopy has two promising applications: testing mass-produced, identical components on a production line, where it can give a low-cost per test and handle production volumes of up to 10,000,000 per year; the other application is in-service monitoring, where a resonance spectroscopy system runs in the background testing a structure continuously and giving an early warning of significant change. It is this second application that will be developed and evaluated in the second generation project with emphasis on concrete structures although the technique should work on steel and wooden structures.
A second meeting will be held in Darmstadt for prospective partners on either 6/7th October or 13/14th October. Please contact David Andrews at CU as soon as possible if you wish to attend (fax: +44 1954 231 494).
Cambridge Ultrasonics (CU) has been developing
ultrasonic transducers for testing concrete since 1986 when the
work was sponsored by the Centre for Marine Technology at Imperial
College London. The present transducers have excellent coupling
to concrete due to the use of a solid coupling agent, which 
also allows the transducers
to work on surfaces of virtually any roughness. They have high
efficiency and a wide frequency range.
In 1986 we realised that the majority of transducers sold for testing concrete were optimised for energy generation and not for wide bandwidth. Our design philosophy at CU was to eliminate as many of the interfaces between the piezoelectric element creating the ultrasound and the concrete surface. To achieve this we created new materials with the same acoustic impedance as concrete - hence the use of fast-setting mortar as a coupling agent.
Theory shows that it is important for any timing system, such as radar, sonar or ultrasonic pulse-echo, to use the bandwidth of the transducer as effectively as possible. We have found that swept frequency chirp signals and matched filtering of the received signal give important benefits in this respect - with better performance than more commonly used pulse excitation. The chirps and matched filters must be carefully designed to suit the transducer. The first successful transducers were tested in 1989 and reported in 1991 at the IEEE symposium on Ultrasonics and Ferroelectrics.
Between 1990 and 1992 we experimented with
a mechanically steerable transmitter with four integrated receivers.
The reasoning was to adapt techniques of mechanically steered
transducers from medical ultrasonics and radar. Unfortunately,
the prototype did not work successfully because of internal 
reverberations in
the transmitter caused partly by the rotating surfaces. Despite
painstaking lapping of the rotating surfaces it proved impossible
to reduce the reverberations to an acceptable level and this research
direction was abandoned. This work was supported by an oil major
and a UK government department.
In 1993 we started working on arrays of transducers and this has provided another breakthrough in performance. Arrays of 7, 16 and 19 transducers have been used. Most of the evaluation experiments have been done using only 7 transducers, with good results. An array allows spatial averaging to be made which helps to suppress the random spatial signal caused by aggregate particles in concrete. The effect is known in medical ultrasound and laser physics as speckle and in radar as clutter or stochastic scattering. Incidentally, we suspect that information about the quality and characteristics of the concrete mix could be extracted from the stochastic signal; in another application it might be desirable to enhance this signal. This stage of development was supported by the German Concrete Association.
The transducers are recommended for research applications and for conventional site testing, where long range is needed. They can be used with conventional pulse-echo systems (with fast-setting mortar as a coupling agent).
The solid coupling agent is a specially formulated fast-setting mortar, which takes a few minutes to reach full working hardness at room temperature (see graph). Modified formulations are available for low temperatures and permanent attachment. A thin metal blade is driven into the mortar to detach the transducer - the mortar is brittle and falls away from both the transducer and test surface. The transducer can be re-used indefinitely.
It takes longer to attach and use this kind
of transducer than a conventional transducer but the greater sensitivity
and range usually allows a test to be made when other transducers
do not work.
After signal processing chirp durations are compressed from 100
us to give sharp pulses of typically 10 
us
- an equivalent axial resolution of approximately 40 mm. This
resolution is close to the size of the largest aggregate particles
used for making concrete and we believe this sets the limit of
what is physically possible. It is possible, therefore, that single
transducers are working as well as they can but the level of performance
is not sufficient for imaging in concrete, the ultimate goal.
That’s why its important to use arrays or synthetic apertures
to get better performance out of a system; synthetic apertures
require mechanical scanning, which is difficult on concrete, and
so arrays offer the best approach.
Range in pulse-echo depends on concrete type but 1.8 m has been achieved so far. The limiting range in transmission has not yet been measured; 2 m has been achieved without difficulty. Generally, the larger the maximum aggregate size the shorter the range. Range values quoted here refer to aggregate sizes of approximately 16 to 25 mm.
Solids support three bulk wave modes: longitudinal compression (scalar wave) and two orthogonal transverse shear (vector waves). Shear waves travel slower than compression and this creates a problem for inspecting concrete that is best illustrated by an example. Consider a single target in concrete scattering ultrasound from an ultrasonic wave. Whatever kind of wave arrives at the target some scattered energy could be converted into another mode; mode conversion is well-known and many shear probes operate using the effect. A receiver is likely to collect both shear and compression waves from the target but since each wave travels at a different speed they will arrive at different times. Timing information is used in all ultrasonic systems to identify the position of scatterers and create images so the mere existence of different wave types can result in phantom targets.
This problem was recognised by CU and our research has resulted in the development of transducers which are able to discriminate between compression and shear waves. Our transducers are optimised for sensitivity to compression and provide about 20 dB suppression of shear. Compression transducers can be more easily used in arrays and are compatible with conventional first time of arrival measurements used by the construction industry. However, we have also made shear transducers that could be useful in detecting cracks caused by the alkali-silica reaction.
Cambridge Ultrasonics (CU) co-ordinated
a project in the European Commission’s Brite/EuRam programme
(BE-5029-92). The objectives were to improve the method of resonance
spectroscopy and to evaluate a range of industrial materials.
The project was successfully completed in 1996. One conclusion
resulting is that resonance spectroscopy can be used to assess
the quality of mass-produced 
components
on a production line.
We have been following developments in the technique at the Los Alamos Laboratory and at Quasar Inc in the United States with interest. Quasar has specialised in using resonance inspection (RI) for rapid testing in high-volume applications. We are pleased to announce that CU has been appointed an agent for Quasar products and services. Our experience in developing and using our own resonance spectroscopy system gives us a strong technical understanding from which to advise and serve potential users of resonance spectroscopy.
CU is now able to offer the following range
of products and services:
Quasar is certified to ISO9000, its products are CE marked for Europe, and have been installed in many locations throughout the world. They are used extensively for testing automotive and aerospace components and users are mainly large industrial organisations. The Quasar's method is protected by international patents.
Shown above is a RI system installed at General Motors capable of handling 10,000,000 components per year. The result of a resonance test is a spectrum. The Quatrosonics RI system collects the amplitude spectrum whereas CU’s system collects both amplitude and phase spectra. Quatrosonics’ data processing depends upon finding the frequency of significant peaks; the criterion for deciding pass or fail depends upon the existence of peaks within appropriately chosen frequency ranges and the possible existence of splitting in peaks. Success-rates for interpretation are commonly close to 100%. CU uses a different approach to data processing. An artificial neural network (multilayer perceptron) analyses the whole amplitude and phase spectra. Success-rates for the neural network on mass-produced metallic components have been in the range 90% to 100%.
The splitting of resonance peaks has been
known for many years to indicate the presence of flaws, particularly
in high 
symmetry
components, where there may be degenerate vibrational modes. The
presence of a flaw usually affects the degenerate modes differently
(degenerate modes have the same vibrational frequency but a different
motion) causing the peaks to take slightly different frequencies,
splitting one peak into two or more peaks.
Electronic engineers have oscilloscopes and logic analysers but what about ultrasonic engineers? One option is to visualize ultrasonic waves (render them visible) - the visualization service from Cambridge Ultrasonics (CU) is a low-cost way to use this specialised technique.
The CU visualizing system uses schlieren or photoelastic optics and our own high speed stroboscope designed specially for this task. Cambridge Ultrasonics was first to develop (1976) and commercialise (1986) the use of modern, solid-state devices for visualizing ultrasound - instead of the spark-gap systems used previously. We offer an investigation service based upon 20 years of experience in this field.
Visualization is the combination of a high-speed stroboscope (light flash duration of 30 ns) and optical contrast methods such as schlieren, shadowgraph or photoelasticity to render visible ultrasonic waves in transparent media (usually air, water or glass). The equipment works with a wide range of ultrasonic systems including: medical ultrasound scanners, conventional pulse-echo inspection equipment, ultrasonic flow-meters, sonar systems and robot guidance systems working in air.
CU’s equipment can be used with most medical scanners to image the beam in real-time, offering a fast alternative to painstaking beam-plotting. Beam-steering can be followed and so too can beam focussing.
A simple example of the power of synchronizing the visible image and other equipment is the case of a receiver transducer and an oscilloscope; by triggering when the stroboscope light flashes all later events shown on the oscilloscope are caused by ultrasound propagating after the flash. By increasing the delay between when the ultrasound is first generated and when the light flashes it is possible to see which waves are just about to enter the receiver - those waves correspond to signals nearest the trigger point. In this way it is possible to identify features on the oscilloscope trace and say, for example: this is the main pulse arriving from the transducer, this is a reflection from the wall of the tank, this is a wave that has been reflected from the bottom of the tank and then reflected from the water surface, this pulse has been stretched because of reverberations in the wall of the tank.
This is the equivalent of a logic analyser for debugging ultrasonic systems because it allows a direct observation of cause and effect. We use the technique extensively on behalf of customers in projects as diverse as short investigations to large development projects.
At CU we believe the limitation of visualization is the ability of the human mind to devise a way of applying the technique.
Are you wrestling with finite element programmes to predict the eigenmodes of gases vibrating in strange shaped containers? Very often a small, unimportant change in experimental procedures can lead to great computational savings - remember Helmholtz and his resonating cavity? In a Helmholtz resonator it doesn’t matter what the shape of the cavity is only its volume is important and that means no finite element geometry to create, no meshing problems, no computer memory shortages and no (ridiculously low) upper frequency limits. It should mean a simple calculation instead of days or weeks of work on a computer.
The frequency of oscillation of a Helmholtz
cavity depends upon how quickly the gas in the neck of the 
cavity
moves in and out, essentially this gas is the mass of an oscillator.
The stiffness in the system is provided by the compressibility
of the gas in the cavity, which depends upon the volume, the gas
and its thermodynamic conditions. The equation for the frequency
of oscillation, f, is simple compared to finite element calculations,
one needs: the speed of sound in the gas, c; the area of the neck,
S; its effective length, L; and the volume of the cavity, V.
This solution assumes the neck is open at one side to the cavity and at the other to an infinite volume of gas, which may not be true. There are solutions available for Helmholtz resonators working into pipes and a range of other cavities. So more complex systems can be solved analytically. Much of the theory can be solved by analogy to electrical circuits so for complex cases it is possible to solve the problem using a SPICE programme instead of a finite element programme - SPICE programmes are less expensive, quick and need no meshing.
If you are using sound to measure the thermodynamic properties of gases try Helmholtz first. A good account of the theory is in Kinsler, E.K., Frey, A.R., “Fundamentals of acoustics” 2nd Ed Wiley 1962.