Des exemples concrets sont donnés au verso de chaque cas :

• Kontrolle der Überprüfbarkeit vor der Herstellung

• Optimierung Ihrer Röntgenprüfungen

• Auswahl und Entwicklung eines Sensors bzw. einer Strahlenquelle

• Beurteilung der Auswirkungen von Verschlechterungsfaktoren

• Ausbildung und Schulung Ihrer Teams

• Validierung eines Prüfverfahrens

• Simulation der Durchführung einer Prüfung

• Optimierung Ihrer Qualifizierungsverfahren

• Steigerung der Zuverlässigkeit Ihrer POD-Kurven

• Steigerung der Diagnosezuverlässigkeit

Les cas d’application sont également disponibles dans les langues suivantes :

• 中文 | English | Español | Français | Italiano | 한국의 | Português

M. Michele Carboni, chercheur et utilisateur CIVA au Département d’Ingénierie Mécanique de l’Université Politecnico di Milano, a aimablement accepté de répondre à nos questions.

Depuis quand êtes-vous un utilisateur CIVA ?

J’ai commencé à travailler avec CIVA sur la version 8.1 en 2006.

Pour quel type d’activités utilisez-vous CIVA ?

En tant que membre du personnel académique du Département d’Ingénierie Mécanique à l’Université Politecnico di Milano (Italie), mon intérêt concerne principalement les sujets de recherche liés aux CND et mon utilisation de CIVA reflète grandement cette situation. Comme je suis également responsable des cours de “Mécanique Expérimentale et CND” pour nos étudiants en Master de Science, certaines applications importantes de CIVA concernent également l’enseignement.

Qu’est-ce qui vous a convaincu d’utiliser la simulation ?

C’est une question intéressante. Au cours des dix dernières années, un besoin de compléter et étayer les résultats expérimentaux avec des simulations numériques a vu le jour au sein de la communauté CND. Ceci est dû au coût parfois très élevé des campagnes expérimentales. J’ai commencé à utiliser et à faire des recherches sur la simulation des CND afin d’apporter ma contribution à ce nouveau courant de pensée.

Qu’est-ce que cela a changé pour vous ?

Cela m’a donné la possibilité d’ajouter à mes recherches et méthodologies habituelles de nouvelles perspectives et de nouveaux outils. Je dois dire que je suis extrêmement satisfait de l’opportunité que m’offre la simulation de comprendre et interpréter certains des résultats expérimentaux que j’obtiens lors de tests. De plus, j’ai également été en mesure d’obtenir de bonnes prédictions de résultats expérimentaux.

Pourriez-vous nous donner une vue d’ensemble des projets de recherche dans lesquels vous êtes impliqué ?

Mon sujet de recherche principal est l’intégrité structurelle. J’ai commencé en 1997, par l’étude et la recherche sur les mécaniques de fatigue et de fracture en tant qu’étudiant en Master de Sciences, et j’ai également continué pendant ma thèse doctorale. En 2006, j’ai commencé à m’intéresser aussi aux CND. J’ai eu la possibilité d’apporter ma participation et de travailler dans le cadre de différents projets régionaux, nationaux (italiens) et européens. La plupart d’entre eux portaient sur l’intégrité structurelle des essieux ferroviaires. Actuellement, je travaille sur des courbes de “Probabilité de Détection” et de leur modélisation (Model-Assisted POD) pour l’inspection ultrasonique des essieux ferroviaires, et sur l’application de l’inspection par les Courants de Foucault pour déterminer la détection des dommages de corrosion-fatigue sur les essieux ferroviaires.

Dans quelles applications la simulation a-t-elle été particulièrement bénéfique à votre travail ?

Elle l’a bien entendu été pour les derniers sujets mentionnés dans la question précédente. L’approche “Probabilité de Détection Assistée par Modèle” est expressément basée sur l’idée de recouper résultats expérimentaux et numériques, aussi les simulations sont-elles d’une importance fondamentale. Je trouve également la simulation très utile pour comprendre le comportement des Courants de Foucault sur un composant à fissures multiples, comme un essieu ferroviaire corrodé par la fatigue.

Quelles fonctionnalités supplémentaires souhaiteriez-vous voir apparaître dans CIVA ?

Plus de flexibilité sur certains aspects des modules POD et ET, car ils n’autorisent parfois pas la représentation du problème physique que je souhaiterais simuler.

ASAP vise à améliorer significativement le contrôle des assemblages de « sécurité » lors de leur réalisation, par le déploiement d’un contrôle non destructif en usine donnant un résultat fiable, rapide, simple, univoque de la santé du point soudé.

Ce projet assurera la mise en œuvre d’une méthode de contrôle ultrasonore multi-éléments innovante et simple d’usage, combinant des développements au niveau du traducteur, du système d’acquisition et des algorithmes temps réel intégrés au système.

Le résultat attendu à la fin du projet est de mettre à disposition du contrôleur un système de contrôle permettant de s’adapter en temps réel à la géométrie externe de la soudure.

A cette fin, les verrous suivants seront levés :

• Développement de capteurs ultrasonores de petite taille au regard de la géométrie à contrôler
• Développement et implémentation temps réel d’algorithmes adaptatifs
• Intégration d’un diagnostic automatisé pertinent pour les contrôles en usine
• Corrélation entre critères CND et tenue mécanique des PSR
• Modélisation macroscopique de prédiction de la tenue mécanique des PSR

ASAP rassemble les partenaires suivants :

Le projet s’étend sur 3 ans (2011-2014) et représente un budget d’investissement de 2026,9k€. Le projet est soutenu par l’ANR (Agence Nationale de la Recherche) à hauteur de 844,6k€.

Motivations derrière ce projet :

Lors d’un accident de la route, la structure du véhicule est conçue pour absorber une partie de l’énergie libérée par le choc, ceci afin de protéger le plus possible les occupants du véhicule. La qualité des assemblages, et plus particulièrement celle des points soudés par résistance, est un enjeu clef pour assurer cette absorption d’énergie par déformation des pièces et rupture de points fusibles.

De plus, dans un contexte concurrentiel toujours plus grand couplé à des défis environnementaux ambitieux, l’industrie automobile doit innover pour proposer des technologies en rupture, mais aussi pour optimiser la conception et la fabrication des véhicules.

L’allègement des véhicules constitue un des leviers permettant de répondre à ces enjeux, sachant qu’une telle orientation technique doit composer avec des performances d’excellence en matière de sécurité automobile.

Le niveau de maîtrise recherché suppose de déployer des démarches d’assurance qualité ambitieuses. Ces démarches restent soumises à la qualité d’un contrôle à la hauteur de cette ambition, et à l’acquisition d’une expertise plus poussée en la matière. Cependant, la mise en œuvre de méthodes d’inspection non destructives reste difficile, puisque les profils de points de soudure présentent une variabilité caractéristique du process industriel.

Cette difficulté se traduit sur le plan opérationnel par des temps de contrôle importants, ainsi qu’un niveau d’expertise et un savoir-faire longs à acquérir, à l’origine d’une disponibilité des compétences lourde à maîtriser.

Les travaux réalisés dans le projet viseront à l’étude et au développement d’une nouvelle méthode de contrôle ultrasonore multi-éléments dédiée aux points de soudure par résistance.

L’une des originalités du projet sera d’implémenter au sein de cette instrumentation des algorithmes innovants, permettant en temps réel de s’adapter à la géométrie complexe de la soudure, d’améliorer les performances de détection par une meilleure focalisation du faisceau ultrasonore, et de disposer d’une imagerie évoluée d’analyse permettant une interprétation immédiate sans besoin d’expertise de la qualité du point soudé.

Une part importante du projet sera par ailleurs consacrée à la caractérisation de la tenue dynamique à posteriori des points soudés, en modes purs et mixtes I/II sur une large gamme de vitesses de déformations, ainsi qu’au développement et à la fiabilisation d’un modèle numérique macroscopique de tenue dynamique des points soudés.

Une autre originalité du projet sera d’établir une corrélation entre modélisation paramétrique, qualité par diagnostic CND et comportement dynamique à rupture des assemblages soudés par points. Le modèle numérique macroscopique, couplé aux outils de simulation de la méthode CND, permettra de valider et d’optimiser la méthodologie de contrôle pour une grande gamme de paramètres caractéristiques des assemblages et points soudés.

L’implication et l’expertise des partenaires d’ASAP dans des domaines de compétences complémentaires permettront d’aboutir, au bout de trois ans, à la création du premier appareil du marché ultrasons multi-éléments assurant un diagnostic fiable, rapide, simple, et univoque de la qualité de la soudure par point, et à la fiabilisation d’un modèle paramétrique de tenue des PSR.

The default arbitrary unit for any computations in the CIVA UT module is the « point » (pts) unit, which corresponds to an absolute unit for the signal amplitude. Besides absolute points units, CIVA systematically provides in the result images (A-scan, B-scan, etc…) the relative amplitudes either in dB or % (the default choice between dB or % can be customized using CIVA preferences menu). This relative amplitude corresponds to the ratio of the current amplitude at the cursor position and the calibration reference amplitude.

It is generally possible to compare point amplitudes coming from different simulation files. However, this absolute comparison must be done carefully so that it makes sense. First, one very important point when the input signal differ from one simulation to the other (for instance when changing the centre frequency), is to ensure that the maximal amplitude of the input signal used in each simulation is the same. When using the synthetic signal, it is advised to keep the default value for this input signal amplitude (100%).

We are going to discuss in this section the validity of amplitude comparisons in following practical cases:

• Inspection configurations using the same transducer
• Inspection configurations using the same transducer but different sampling of the signal
• Inspection configurations using transducers at same frequency and same thickness but with different apertures
• Inspection configurations using transducers at different frequencies
• Inspection configurations using different transducers (different thicknesses) at same frequency

NB: the term transducer refers to the piezoelectric element and consequently « using the same transducer » induces no change in the piezoelectric material and in its shape (all the dimensions of the crystal). It is recalled that the crystal thickness is directly linked to its resonance frequency and so to the centre frequency of the corresponding probe.

Comparison between two different inspection configurations using the same transducer (same frequency, aperture, etc.)

It is possible to do a direct comparison in this case between absolute amplitudes.
Simulation allows evaluating with a good precision the influence of different inspection settings (different incidence angles, different material properties or coupling conditions, etc…).

Comparison between two different inspection configurations using the same transducer but different sampling of the input signal (different sampling frequency and/or number of samples)

It is also possible to do a direct comparison in this case between absolute amplitudes, as soon as the signal sampling is in agreement with the CIVA rules.

Comparison between two different inspection configurations using two transducers at same frequency and same thickness but with different apertures (different crystal shapes and dimensions)

It is also possible to do a direct comparison in this case between absolute amplitudes.
Indeed, we can reasonably consider in a good approximation that the electro-acoustic transduction and the particle velocity don’t vary with respect to the crystal dimensions (for instance, the diameter for a circular probe). The model assumes that the transducer acts as a piston: the velocity is uniform on its surface, whatever the size and the shape of the radiating surface.

Comparison between two different inspection configurations using transducers at different frequencies

It is not possible to do a direct comparison in this case between absolute amplitudes. The reason is that the change in frequency has an influence on the electro-acoustic transduction (which is not accounted for in CIVA) and not only on the wave propagation and defect interaction. However, the comparison between these different configurations can be done relatively, which means by studying the normalized amplitude variation of the echoes versus frequency (i.e ratio of the echoes amplitudes between a defect and the calibration flaw). This is anyway not different from what is done in a real inspection where the calibration is performed for each transducer.

In order to compare the frequency variation observed on simulated echoes regarding equivalent measurements, the CIVA user needs to:

• Do measurements at each frequency: one on the current defect and one on the calibration reflector.
• Perform the simulations of these two experiments at each frequency, using in modeling a relevant input signal for each frequency.
• Normalize at each frequency the echoes amplitudes by those of the reference defect both on experimental and simulated data.

When selecting a calibration flaw, make sure to respect the criteria defined here.

NB: in presence of attenuation for one frequency of interest, the operation described in the section concerning the choice of the input signal must be considered.

Comparison between two different inspection configurations using different transducers (different thicknesses) at same frequency

Due to different thicknesses, it is not possible to do a direct comparison in this case between absolute amplitudes. The reason and the method to bypass this is the same as for comparison at different frequencies. You can refer to the previous section « Inspection configurations using transducers at different frequencies » for more details.


Back to the UT Module Validation menu
Back to the Validation menu

In order to reproduce the experimental inspections, the user has to define in CIVA the input signal for the « Defect Response » module. In addition, as the electro-acoustic transduction is not modeled by the software, a calibration stage is necessary in order to compare the CIVA simulated results with experimental results in terms of signal amplitudes.

This page deals with:

• The input signal in Civa
• The reference amplitude
• The calibration flaw

You will also find in this page some « pratical information » for Civa users about input signal, sampling of the input signal and reference for the calibration amplitude.


Back to the UT Module Validation menu
Back to the Validation menu

Input signal in CIVA

In the Civa “Defect Response” module, the input signal theoretically corresponds to the second time derivative of the acoustic particle velocity normal to the crystal surface (detailed information explaining this relation can be asked to the support team).
But, as this acoustic particle velocity is not easy to determine, we propose below simple ways to obtain this input signal.

In CIVA, the input signal can either be loaded from an external text file (obtained from a measured calibration echo) or defined in CIVA as a “synthetic” signal (defined with parameters assuming a Hanning or a Gaussian frequency distribution).

• In the case of a synthetic input signal, the centre frequency, bandwidth and phase of the input signal have to be determine by the user, and we will propose detailed methods for their determination.
• In the case of an experimental input signal (external text file), we will explain which calibration flaws are used to measure the calibration echo.

This will be illustrated on an example: the determination of the input signal for a SV45° inspection performed with an immersion probe (planar, Ø6.35mm, 2.24MHz, water path 25mm).

Practical information about input signal
To access the input signal in Civa, you have to open the « Signal » tab from the « Probe » panel.

• If only manufacturer data are available or if a synthetic signal is going to be used as input signal, you click on the « edit » button and a new window appears where you can define the waveform, the phase and the sampling of the input signal.
• If a specular echo from a reference flaw is stored as a text file, it should be loaded by clicking on the « load » button.

Synthetic input signal

The aim of this section is to answer this question:
How to determine the centre frequency, bandwidth and phase of the input signal?

Two method will be explained for the frequency and the bandwidth. Then some information will be given about the phase and the sampling of the input signal:

• First method: from data provided by the probes manufacturer
• Second method: from the experimental echo of a calibration flaw
• How to determine the phase of the input signal?
• Practical information about the sampling of the input signal

First method: from data provided by the probes manufacturer
If available, the probe parameters from the manufacturers (centre frequency (fc), bandwith (BW)) are often sufficient to define a reference signal in terms of waveform. The phase, in this case, is put to 0° for example.

Second method: from the experimental echo of a calibration flaw
The centre frequency and bandwidth of the input signal can be directly deduced from the experimental echo of a calibration flaw.

In our example, this experimental echo is the specular echo of a Ø2mm Side Drilled hole (SDH) positioned at 4mm depth. The Fscan of this echo is used to determine the centre frequency and bandwidth of the input signal:

(Left) Ascan of the echo of the Ø2mm SDH at 4mm depth, (Middle and Right) Fscan of this specular echo

The centre frequency of the input signal is 2.24MHz. The bandwidth of the input signal is 61%.

As we can see on the next figure, the measured and Civa simulated amplitudes of SDH at different depths are in good agreement (discrepancies of less than 2 dB).

Comparison of measured (black) and simulated (red) responses of SDH Ø2 mm at different depths from 4 mm to 60 mm (step 4mm).
Input signal: fc=2.24MHz, BW=61% and phase=0°.

How to determine the phase of the input signal?
The inspection often requires only the envelop of the input signal. But if the phase is necessary to enable the comparison between the experimental and simulated Ascans, it can be adjusted: the procedure consists in adjusting the phase parameter of the input signal in order to reproduce after simulation of the reference flaw inspection the same signal as the experimental echo from the reference flaw.
For further information of the phase obtained after simulation on a SDH or a FBH, you could contact the support team.

In our example, the reference flaw is a Ø2mm Side Drilled hole (SDH) positioned at 4mm depth. We calculated with Civa the echoes of this reference SDH with different input signals having the previously determined centre frequency (2.24MHz) and bandwidth (61%) but having various phases (0°, 280°, 300° and 320°, see figures below). We compare the different simulated Ascans obtained with the experimental one (figure below).

Superposition of measured (black) and Civa simulated (red) Ascans of the echoes of the Ø2mm SDH at 4mm depth obtained for 4 different phases of the input signal (0°, 280°, 300° and 320°).

We can see that the simulated Ascan obtained with the phase 300° is very close to the experimental one: this value of 300° will be chosen for the input signal.

As we can see on the next figure, with this input signal (fc=2.24MHz, BW=61% and phase=300°), the measured and Civa simulated amplitudes of SDH at different depths are still in good agreement (as previously said and noticed by comparing figures corresponding to different phases, the phase of the input signal doesn’t have a strong effect on the Ø2mm SDH echo amplitudes, the phase adjustment is useful to compare measured and simulated Ascans).

Comparison of measured (black) and simulated (red) responses of SDH Ø2 mm at different depths from 4 mm to 60 mm (step 4mm).
Input signal: fc=2.24MHz, BW=61% and phase=300°

Practical information about the sampling of the input signal
The input signal has to be digitized. The sampling frequency of the input signal should be at least 25 times its centre frequency. The number of points just needs to be large enough so that the signal at starting and ending times is close to zero, as in the following image.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Experimental input signal

In Civa “Defect Response” module, as previously said, the input signal is supposed to be the second time derivative of the acoustic particle velocity. It can be shown (contact the support team for additional information) that in most inspection cases the correct input signal can be deduced in a good approximation from the echo measured on a calibration reflector.

In our example, we choose as an experimental input signal, the specular echo of a Ø2mm Side Drilled hole (SDH) positioned at 4mm depth. This specular echo is represented on the next figure (only the specular contribution of the echo is used as input signal, the creeping wave contribution has to be eliminated):

(Left) Measured echo of the Ø2mm SDH at 4mm depth, (Right) Specular contribution of this echo (part which will be used used as input signal).

As we can see on the next figure, with this experimental input signal, the measured and Civa simulated amplitudes of SDH at different depths are still in good agreement.

Comparison of measured (black) and simulated (red) responses of SDH Ø2 mm at different depths from 4 mm to 60 mm (step 4mm).
Experimental input signal (SDH Ø2 mm at 4mm depth echo)

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Reference Amplitude

To compare measured and Civa simulated amplitudes, the user has to assign a reference amplitude which is the output echo signal obtained from a calibration flaw, as it is the case in real inspection procedures. Indeed, for the Defect Response module, it is needed to make a calibration measurement on a reference flaw in order to take into account the electro-acoustic transduction which is not modeled in CIVA. This calibration is used to give a physical sense to the CIVA simulated echo amplitudes. The absolute amplitude of the CIVA simulated echo does not represent the amplitude of the received echo since the electro-acoustic transduction is not modelled. That’s why, the amplitudes simulated with the CIVA Defect Response module have to be considered relatively to that of the calibration flaw: the user has to consider for a given inspected flaw a “relative amplitude” which is the ratio of the current flaw echo amplitude to the amplitude of the reference flaw echo.
This relative amplitude physically corresponds to the ratio of the received electrical signals for the current and calibration defects. The value of this ratio is consequently experimentally measurable and can lead to comparisons between simulation and experiment. In that goal, the user has to evaluate and compare the relative amplitude obtained for a given flaw both in simulation and in measurement.
The use of this amplitude normalization by a calibration technique allows to overcome the complex modelling of the electro-acoustic transduction (not modelled in CIVA), phenomenon already occurred in the echo on the calibration flaw.
Ask the support team for more information about this calibration.

Practical information about Reference Amplitude
In the Defect response module, CIVA offers 3 ways for defining the 0dB reference amplitude in the « Calibration » tab from the « Computation parameters » panel:

• None: the default reference is the strongest echo of the simulation. Then, the calibration flaw can be directly defined in the simulation configuration to give a reference and allow direct comparison. But the 0dB value may not be assigned to this echo but a stronger one.
• Manual: CIVA asks for a value in points, the CIVA arbitrary unit of echoes absolute amplitude. In this case, it is advised to previously run a simulation with the reference flaw, to note the value in points of the corresponding simulated amplitude (from an Ascan or from the “information” menu) and then to put this value in the calibration tab to be used as the 0dB reference for the others simulation.
• Simulation: CIVA offers the possibility to run a pre-simulation to compute the amplitude of the echo from the chosen flaw in the chosen calibration block and then automatically uses this as the 0dB reference. In Civa 10, this option is only available for mono-element probes.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Calibration flaw

In order to provide a reliable reference signal, the flaw used as reference for calibration has to:

• interact with the beam in a specular or pseudo-specular way
• be located in the far field if using a non focused probe
• be located in the focal zone or further if using a focused transducer
• be lighted by the beam during a bounded time
• be large enough (wavenumber.radius=k.a>1.5 for a FBH , 1.5<k.a<20 for a SDH).

As soon as they respect this criteria and provided that the corresponding signals are also considered as reliable in the real inspection setup (repeatability, etc.), different type of defect (SDH, FBH, flat surface…) and different ultrasonic modes (P0°, SV45°…) can be considered. In Civa defect response, typically, Ø2 or Ø3mm SDH reflectors can be used then usually 1.5<k.a<2 as soon as the frequency is higher than 1MHz.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Generalities

In all cases, the tests are usually performed with probes radiating SV45° or P45° waves upon planar blocks containing vertical backwall breaking flaws. A scanning along X and Y axes is applied in these inspections.

The maximum amplitude of the rectified echoes and the C-scan are stored. Only the maximum amplitude from the corner echo will be extracted.

Back to the UT Module Validation menu
Back to the Validation menu

Several planar specimens are used in the followings experiments. The aim is to evaluate the influence of different parameters.

• Analysis of the echoes from a single notch inspected with P waves generated by a contact probe
• Notches of different heights inspected with SV45° waves generated by contact probes(1 extension)
• Notches of different heights inspected with SV45° waves generated by contact probes (2 extensions)
• Notches of different extensions inspected with SV45° waves generated by immersion probes
• Notches of different extensions inspected with P45° waves generated by immersion probes
• Notches of different extensions inspected with P60° waves generated by immersion probes
• Notches of different dimensions inspected with SV45° waves generated by immersion probes

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Analysis of the echoes from a single notch inspected with P waves generated by a contact probe

Configuration

In order to evaluate all the echoes reflecting from a backwall breaking notch with a contact probe radiating P waves, this experiment is performed on a 37mm high planar stainless steel specimen containing a 10mm high and 20mm width backwall breaking notch.

The experiments involve a contact probe. For this Ø12.7mm circular contact probe, the P45° mode is used for inspection. The input signal frequency is 2.25MHz, with 50% bandwidth and 280° phase.

Results

The Bscan are recorded from simulated and experimental data:

The corner echoes from P waves, from SV waves and from converted waves are observed. Sometimes, conversion mode corner echoes are also called mixed corner echoes. Direct and indirect edge diffraction echoes are also identified.

The Bscan can be reconstructed along the P or SV mode ray pathes:

The results are calibrated versus a Ø2mm SDH at 37mm depth.

A good experiment/simulation agreement is obtained for the P, SV and conversion mode corner echoes amplitudes.

The SV mode corner echo is a little bit overestimated and shows two main contributions in the B-scan. It has been further investigated.

Click here for more information.

The main contribution is the P-defect-P-backwall-SV corner echo, as expected with a very favourable path and a strong SV-backwall-P field amplitude.
The SV-defect-SV-backwall-P corner echo contributes with 8dB lesser than the main contribution.


The paths of the contributions are shown here:





The beam corresponding to each mode have been individually computed including one reflection on the backwall (bw):





The simulated SV corner echo is divided into 2 contributions, the second one being not observed experimentally. The SV beam splitting is due to limitations of the model that occur in this case. The strong variation of the transmission coefficient close to the critical angle splits the SV incident beam and consequently the SV reflected beam, which gives rise to a non realistic secondary SV corner echo. However it can be noticed that it has no effect on the main SV corner echo as it can be seen that the discrepancy is only 2dB and the echo is well located.


The Civa development team is working on this specific problem that occurs around the critical angle in order to improve the models already implemented in CIVA.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different heights inspected with SV45° waves generated by contact probes (1 extension)

Configuration

Then, the case of notches of different heights with the same extension and depth is carried out. The specimens are 30mm high planar blocks containing a backwall breaking notch with a 40mm extension. In the first block the notch is 10mm high, and it is 3.2mm high in the second block.

A Ø6.35mm circular mono-element contact probe is used. The SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 44% bandwidth and 147° phase.

Results

The results are calibrated versus a Ø2mm SDH of 40mm extension at 28mm depth.

The comparison shows a good agreement, the maximum discrepancy being 1.5dB.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different heights inspected with SV45° waves generated by contact probes (2 extensions)

Configuration

In this case, the notches of several heights are inspected for 2 different extensions. The mock-up is a planar stainless steel block containing backwall breaking notches of height 2mm, 5mm, 10mm and 20mm for extension 40mm and 5mm. An additional Ø2mm SDH with an extension of 60mm is located at 30mm depth for calibration.

Two mono-element contact probes are used:

The results corresponding to each probe are available by clicking on the size of the probe.

Results

For the 20*22mm rectangular contact probe at 2.0MHz, the SV45° mode is used for inspection. The input signal frequency is 2.0MHz, with 41% bandwidth and 75° phase. The results are calibrated versus the Ø2mm SDH at 30mm depth.

There is a good agreement.

For the Ø6.35mm circular contact probe at 2.25MHz, the SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 44% bandwidth and 147° phase. The results are calibrated versus the Ø2mm SDH at 30mm depth.

There is an overall good agreement.

The curves show a good agreement with discrepencies lower than 2dB, except for the small notch with the circular probe. In this case the maximum discrepancy is 4dB. As the flaw size (2mm) is close to the wavelength (λ = 1.6mm), this is coherent with well known limitations of the Kirchhoff approximation used in the CIVA response which is a high frequency approximation.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different extensions inspected with SV45° waves generated by immersion probes

Configuration

In the following experiment, the mock-up is a 20mm high planar steel block with 4 backwall breaking notches: 3 notches of height 2mm and extension 2mm, 5mm and 15mm, and a notch of height 6mm and extension 15mm.

Two immersion probes are used:

Results

The notches are associated with a number in order to display the results:

For the Ø6.35mm circular immersion probe at 2.25MHz with 20mm water path, the SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 64% bandwidth and 290° phase.
The results are calibrated versus a Ø2mm SDH of 40mm extension at 20mm depth.

There is a good agreement with less than 2dB discrepancy.

For the Ø12.7mm circular immersion probe at 4.5MHz with 20mm water path, the SV45° mode is used for inspection. The input signal frequency is 4.5MHz, with 73% bandwidth and 270° phase.
The results are calibrated versus a Ø2mm SDH of 40mm extension at 15mm depth.

There is a very good agreement with less than 0.5dB discrepancy.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different extensions inspected with P45° waves generated by immersion probes

Configuration

In order to evaluate the echoes reflecting from a backwall breaking notch with an immersion probe radiating P45° waves, this experiment is performed on a 20mm high planar steel block with 4 backwall breaking notches: 3 notches of height 2mm and extension 2mm, 5mm and 15mm, and a notch of height 6mm and extension 15mm.

The experiments involve an immersion probe. For this Ø6.35mm circular immersion probe with 25mm water path, the P45° mode is used for inspection. The input signal frequency is 4.7MHz, with 56% bandwidth and 255° phase.

Results

The Bscan are recorded from simulated and experimental data. They are also reconstructed using P or SV mode ray paths. This example comes from the inspection of the 6mm high and 15mm width vertical backwall breaking flaw.

The corner echoes from P45° waves, from SV waves and from converted waves are observed. Direct and indirect edge diffraction echoes are also identified.

The notches are associated with a number in order to display the results:

The results are calibrated versus a Ø2mm SDH at 5mm depth.

A good experiment/simulation agreement is obtained for the P and SV corner echoes amplitudes.

The observation of a split simulated SV corner echo made for the analysis of corner echoes regarding the division in two contributions is still valid. This splitting of the SV corner echo in simulation is observed for all the notches. As for the previous probe, the splitting is present in the SV direct beam and in the SV beam reflected beam on backwall (bw).

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different extensions inspected with P60° waves generated by immersion probes

Configuration

In order to evaluate all the echoes reflecting from a backwall breaking notch with an immersion probe radiating P60° waves, this experiment is performed on a 20mm high planar steel block with 4 backwall breaking notches: 3 notches of height 2mm and extension 2mm, 5mm and 15mm, and a notch of height 6mm and extension 15mm.

The experiments involve an immersion probe. For this Ø6.35mm circular immersion probe with 25mm water path, the P60° mode is used for inspection. The input signal frequency is 4.7MHz, with 56% bandwidth and 255° phase.

Results

The Bscan are recorded from simulated and experimental data. This example comes from the inspection of the 6mm high and 15mm width vertical backwall breaking flaw.

The corner echoes from P60° waves, from SV waves and from converted waves are observed as for the P45° radiating probe. Direct and indirect edge diffraction echoes are also identified.

The results are calibrated versus a Ø2mm SDH at 5mm depth.

For all notches, a very good experience/simulation agreement is obtained for the P corner echo amplitude. For the conversion mode corner echo there is also a good agreement with less than 3dB discrepancy.
But some important discrepancies (up to 12dB) are observed for the SV corner echo over all notches. This phenomenon is likely to be similar to the one for the P45° waves discussed above. This configuration is more sensitive since the SV beam angle is closer to the critical angle.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Notches of different dimensions inspected with SV45° waves generated by immersion probes

Configuration

The next experiment evaluates the influence of the notch height for several extensions. The 5mm high specimen contains backwall breaking notches of height 0.5mm, 1.5mm and 2.5mm for 2mm, 5mm and 15mm extensions.

The measurements are performed with two Ø6.35mm circular immersion probes.

Results

For the Ø6.35mm circular immersion probe at 2.25MHz with 20mm water path, the SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 64% bandwidth and 290° phase. The results are calibrated versus a Ø2mm SDH of 40mm extension at 4mm depth.

For the Ø6.35mm circular immersion probe at 4.7MHz with 10mm water path, the SV45° mode is used for inspection. The input signal frequency is 4.7MHz, with 56% bandwidth and 255° phase. The results are calibrated versus a Ø2mm SDH of 40mm extension at 15mm depth.

The agreement is very good for all cases with the 5Mhz probe and for the 2 higher flaws with the 2.25 Mhz probe (less than 2dB discrepancies).
For smaller notches at 2.25 MHz, we observe a stronger difference between measured and simulated amplitudes but this difference remains acceptable (less than 4dB difference). As the size of the flaw gets closer to the wavelength, it can be expected in these cases to obtain less precised results.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Conclusion for corner echoes

A very good agreement is obtained between measured and simulated results for corner echoes from SV45° waves: the gap between the measured and simulated corner echo maximum amplitudes is less than 2 dB for most of the configurations studied.

In few cases, some discrepancies (until 4 dB) are observed for small notches relatively to the probe center frequency. They occur for example in the following cases:

• Measured/simulated discrepancies between 2 and 4 dB are observed for the notches of 2 mm height with the contact probe at 2.25 Mhz (wave length λ = 1.4mm) but not for the notches of greater heights
• Measured/simulated discrepancies of about 4 dB are observed for the smallest notches (0.5 mm height) with the immersion probe at 2.25 MHz (wave length λ = 1.4mm). The discrepancy between simulation and experiment increases as the flaw size is smaller than the wavelength. This is coherent with well known limitations of the Kirchhoff approximation used in the CIVA response which is a high frequency approximation.

A very good agreement is also obtained between measured and simulated results for corner echoes from P waves: the gap between the measured and simulated corner echo maximum amplitudes is less than 2 dB for most of the configurations studied.

The corner echo from the SV waves associated with P waves is sometimes badly estimated by Civa when incidence is close to the critical angle. This is due to a splitting in the simulated SV beam as explained above.

Further studies will be carried out on this topic.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

In this part we consider echoes generated by Flat Bottom Holes (FBH) at different depths with different contact probes. The interaction between beam and FBH is simulated with the Kirchhoff model.

Then, the echoes from SDH and FBH have been compared at specular incidence, using SOV interaction model for SDH and Kirchhoff interaction model for FBH.

Flat Bottom Holes

While investigating Flat Bottom Holes (FBH) responses, simulated data is compared to Krautkrämer DGS curves in a first part and to experimental data from 45° tilted FBH in a second part.

DGS Curves

Global overview:

Configuration

DGS (Distance Gain Size) curves from manufacturer (Krautkrämer) are compared to simulation data for some probes. For each FBH diameter, this steel specimen with 11 FBH at different depths has been modeled:

For each hole the maximum amplitude of the specular echoes of the FBH is measured relatively to a calibration hole. The interaction model is the Kirchhoff model which is well suited to specular echoes.

The measurements have been made with the following contact probes:

The results corresponding to each probe are available by clicking on the size of the probe.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Results

Mono-element contact probe 2.0MHz, 20*22mm, SV45°

For the 20*22mm contact probe at 2MHz, the SV45° mode is used for inspection. The input signal frequency is 2.0MHz, with 41% bandwidth and 75° phase.
The results are calibrated versus the Ø3mm FBH at 100mm depth.

There is an overall good agreement with often less than a 2dB discrepancy.

Mono-element contact probe 2.0MHz, 20*22mm, SV60°

For the 20*22mm contact probe at 2MHz, the SV60° mode is used for inspection. The input signal frequency is 2.0MHz, with 40% bandwidth and 0° phase.
The acoustic focusing depth is 36mm, deduced from the simulated beam as illustrated below, which corresponds to a 63mm distance.

The results are calibrated versus the Ø1.5mm FBH at 200mm distance at -44.5dB.

The curves show a good agreement between simulated and experimental data for FBH deeper than the focal depth. For FBH which depth is smaller than the acoustic focusing depth, CIVA over-estimates the echo: for example, for Ø6mm, Ø8mm or Ø12mm FBH, the discrepancy is around 5dB.

Mono-element contact probe 2.0MHz, 24mm, P0°

For the 24mm contact probe at 2MHz, the P0° mode is used for inspection. The input signal frequency is 2.0MHz, with 59% bandwidth and 0° phase.
The acoustic focusing depth is 51mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus an infinite reflector at 100mm depth.

A very good agreement is obtained in P0° mode inspection. CIVA estimates the echo from FBH with less than 1dB difference with Krautkramer data. It can be noted than all the FBH of this experiment are deeper than the focal spot.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Mono-element contact probe 4.0MHz, 8*9mm, SV60°

For the 8*9mm contact probe at 4MHz, the SV60° mode is used for inspection. The input signal frequency is 4.0MHz, with 42% bandwidth and 151° phase.
The acoustic focusing depth is 12mm, deduced from the simulated beam as illustrated below, which corresponds to a 21mm distance.

The results are calibrated versus the Ø0.5mm FBH at 30mm distance at -39dB.

CIVA simulation data are close to Krautkramer data with less than 2dB difference for FBH deeper than the focal depth. For FBH no deeper than the focal depth, CIVA over-estimates the echoes by 2dB for small (Ø0.5mm) FBH and large (Ø10mm) FBH and up to 8dB for medium (Ø4mm) FBH.

Conclusion

For each FBH diameter, there is a very good agreement for the FBHs echoes amplitudes in the far field where the specular echo amplitude linearly decreases with the depth on the figures, but there are discrepancies for the highest diameters at the smallest depths.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Experimental data from 45° tilted FBH

Global overview:

Configuration

Some measurements have been carried out in SV45° and P45° modes upon a planar surface block containing a series of 45° tilted FBH (Ø:1mm, 3mm and 6mm) at different depths (from 5mm to 60mm with a step of 5mm and then at 80mm, 100mm, 125mm and 150mm). The specimen and flaws are represented in the following figure.

For each probe an inspection of the surface of the block is realized and a C-scan obtained. In the C-scan image, the backwall echo has been removed in order to enhance the FBH echoes. However the B-scan extracted on one line of holes shows the backwall echo and the FBH echoes.

For each hole the maximal amplitude of the specular echoes of the FBH is measured relatively to a calibration hole. The interaction model is the Kirchhoff model which is adapted to specular echoes.

The measurements have been made with the following contact probes:

The results corresponding to each probe are available by clicking on the size of the probe.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Results

The maximal amplitude of the specular echoes of the FBH are estimated relatively to a calibration reflector. The DGS curves are displayed for all the probes from simulated, measured or from litterature data in the next figures.
The results show a good agreement in most cases:

Mono-element contact probe 2.0MHz, 20*22mm, SV45°

For the 20*22mm contact probe at 2MHz, the SV45° mode is used for inspection. The input signal frequency is 2.0MHz, with 41% bandwidth and 75° phase.
The results are calibrated versus the Ø3mm FBH at 80mm depth.

There is an overall good agreement with often less than a 2dB discrepancy. For smaller holes at smaller depths, Civa slightly underestimated the amplitude of the echoes. Measured curves are less smooth than simulated curves, but show however a good agreement.

Mono-element contact probe 2.0MHz, Ø12.7mm

For the Ø12.7mm circular contact probe at 2.0MHz, the SV45° mode is used for inspection. The input signal is the inverse P45° experimental direct specular echo of a FBH F3mm, tilt 45° at 30mm depth.
The results are calibrated versus the Ø2mm SDH at 20mm depth in another block inspected with SV45° waves.

Even if the curves from Civa are smoother,there is a good agreement between measure and simulation for Ø1mm FBH and Ø6mm FBH. Civa overestimated the echo from the deep Ø3mm FBH, from 2dB at 60mm to 6dB at 150mm.

In addition to the good agreement in amplitude between simulated data and experimental data, the waveforms also show a very good agreement.
The following curves superimpose the direct specular echo from Ø6mm FBH at different depths with this contact Ø12.7mm probe at 2MHz generating refracted SV45° waves from experimental (black) and simulated (red) data.

Mono-element contact probe 2.25MHz, Ø12.7mm

For the Ø12.7mm circular contact probe at 2MHz, the P45° mode is used for inspection. The input signal frequency is 2.25MHz, with 50% bandwidth and 280° phase.
The results are calibrated versus the Ø2mm SDH at 8mm depth in another block inspected with SV45° waves.

The curves show a good agreement between measure and simulation with less than 2dB difference from the 3 FBH sizes. Experimental curves are less smooth than the curves from simulation data.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Conclusion

An overall good agreement is observed between simulated results using Kirchhoff interaction model and experimental measurements. A very good agreement is also obtained in P0° mode inspection for the infinite reflector. However, some discrepancies with experimental results are observed in the near field.

It can also be seen that in some cases the measured echo-dynamic curves from FBH show irregularities. Those irregularities in the FBH measured curves are due to some anisotropy in the steel of which is made the specimen. Preliminary tests (not presented here) show that these irregularities increase with the frequency.

A simulation based study of the effect of non perfectly planar surface of the FBH, depending on the probe’s frequency, is in progress.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Comparison Side Drilled Holes and Flat Bottom Holes

Configuration

In order to perform this comparison a 50mm thick mock-up contained three SDH (Ø2mm, Ø1.5mm, Ø1mm) and one FBH (Ø3mm tilt 45°) at 30mm depth.

• Amplitude analysis: for a given probe, the measured and simulated maximum amplitude of the specular echo of the three SDH and of the FBH are compared. The FBH is used for amplitude calibration. The aim of these measurements is to perform an experimental validation of the SDH simulation by using a FBH as calibration reflector.
• A-scan analysis: in some cases, the A-scans are also stored, which enlightens the creeping wave around the SDH.

The measurements are performed with contact mono-element and phased-array probes in pulse echo mode. The parameters of the probes are given in the following tables.

Mono-element probes:

Phased-Array probe:

The results corresponding to each probe are available by clicking on the size of the probe.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Amplitude analysis

Mono-element contact probe 2.25MHz, Ø6.35mm

For the Ø6.35mm contact probe at 2.25MHz, the SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 44% bandwidth and 147° phase.
The acoustic focusing depth is 3mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø3mm FBH at 30mm depth.

Mono-element contact probe 2.25MHz, Ø12.7mm, P45°

For the Ø12.7mm contact probe at 2.25MHz, the P45° mode is used for inspection. The input signal frequency is 2.25MHz, with 50% bandwidth and 280° phase.
The acoustic focusing depth is 8mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø3mm FBH at 30mm depth.

Mono-element contact probe 2.25MHz, Ø12.7mm, SV45°

For the Ø12.7mm contact probe at 2MHz, the SV45° mode is used for inspection. The input signal is the inverse P45° experimental direct specular echo of a Ø3mm FBH, tilt 45° at 30mm depth.

The acoustic focusing depth is 17mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø3mm FBH at 30mm depth.

Phased array contact probe of 20 elements with 0.7mm pitch

For the phased array contact probe of 20 active elements (out of 48) with 0.7mm pitch at 5MHz, the P45° mode is used for inspection. The focal law is a P45° beam steering.
The acoustic focusing depth is 20mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø3mm FBH at 30mm depth.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

A-scan analysis

An additional study is carried out concerning the A-scans of the signals with one contact probe and the phased-array probe with the previous configuration.

Mono-element contact probe 2.25MHz, Ø6.35mm

For the Ø6.35mm contact probe, the echoes from all reflectors are well estimated.

All the A-scans show a good agreement between experiment and simulation.

Phased array contact probe of 20 elements with 0.7mm pitch at 5MHz

For the contact phased-array probe, the echoes from all reflectors are well estimated.

The A-scans of the specular echoes generated by the holes are well estimated by Civa.
Moreover, the A-scans enlighten the fact that a creeping wave is also present for associated SV mode echo. This creeping wave propagates around the circumference of a SDH and create an additional echo, which is also correctly predicted.

Conclusion

A good agreement is obtained between the measured and simulated amplitudes of the specular responses of SDH of different diameters relatively to the specular response of a FBH (discrepancy less than 1dB).


Depending on the probe and the mode (P or SV), the experimental direct echoes of the three SDH are stronger or weaker than the experimental FBH direct echo. This relation is well reproduced in simulation. In most cases, the shapes of the simulated FBH and SDH echoes are very close to the experimental one for both SV45° and P45° modes (once the input signal is well calibrated).


The creeping wave (see the SDH Ascans obtained for SV modes) is well simulated in all cases (good shape, position in time and amplitude relatively to the specular contribution), although some discrepancies occur on the smallest SDH.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Side Drilled Holes at different depths

In this part we consider echoes from Ø2mm Side Drilled Holes (SDH) at different depths with different probes:

• Mono-element Immersion probes
• Mono-element Contact probes
• Phased-Array Contact probes

The results show a very good agreement. It can be noticed that Civa generally underestimates the amplitude of the echoes in the very near field (less than 4dB discrepancy).


Back to the UT Module Validation menu
Back to the Validation menu

Side Drilled Holes at different depths and Immersion probes

Global overview:

Configuration

This validation experiment deals with Ø 2mm SDH at different depths. The measurements are performed upon a planar steel block containing Ø2mm SDH from 4 to 60mm depth with 4mm steps. As a reminder, the steel parameters are: density 7.9, P waves velocity: 5900m/s and SV waves velocity: 3230m/s. Since the SDH are inspected perpendicularly to their axis, the SOV interaction model is considered.


The following picture presents the mock-up that is used.

After scanning along the surface of the block, the following B-scan is displayed:

In this configuration, 6 different immersion probes have been used in Pulse Echo mode. All the probes are circular and have a flat surface (none focused):

The results corresponding to each probe are available by clicking on the size of the probe.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Results

For each probe, the simulated P or/and SV beam radiated in the specimen and in the incidence plane is displayed for at least one configuration.
The superimposition of measured and simulated curves of maximal relative amplitude of the P and SV specular echoes of the SDH versus the SDH depths are presented in the next figures.

Mono-element immersion probe 2.0MHz, Ø19mm

For the Ø19mm circular immersion probe at 2MHz with 50mm water path, the P0° mode is used for inspection. The input signal frequency is 2.0MHz, with 60% bandwidth and 0° phase.

The acoustic focusing depth is 19mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 20mm depth.

There is a good agreement between the results from the measurements and the results from CIVA software. The discrepancy is always less than 1dB.

Mono-element immersion probe 2.25MHz, Ø12.7mm

For the Ø12.7mm circular immersion probe at 2.25MHz with 50mm water path, the P0° mode is used for inspection. The input signal frequency is 2.25MHz, with 60% bandwidth and 0° phase.

The acoustic simulated beam is illustrated below.

The results are calibrated versus the Ø2mm SDH at 12mm depth.

There is a good agreement between the results from the measurements and the results from CIVA software.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Mono-element immersion probe 2.25MHz, Ø6.35mm

For the Ø6.35mm circular immersion probe at 2.25MHz, the P0° mode is used for inspection with 20mm waterpath, and the SV45° with20mm water path. An experimental input signal (the SV45° specular echo of a Ø3mm and 45° tilted Flat Bottom Hole at 10 mm depth) is used for P0° waves. The input signal frequency is 2.25MHz, with 64% bandwidth and 290° phase for the SV45° wave.
The acoustic simulated beam is illustrated below for the P0° mode.

The results are calibrated versus the Ø2mm SDH at 8mm depth.

There is a good agreement between the results from the measurements and the results from CIVA software. The maximum discrepancy is 1dB for P0° and 2dB for SV45°.

Mono-element immersion probe 2.4MHz, Ø20mm

For the Ø20mm circular immersion probe at 2.4MHz with 20mm water path, experiments are carried out with SV45°, SV60°, P45°, P60° and P0° modes. It can be noticed that for P45° and P60°, an associated SV beam radiates at 22° and 26° respectively. The input signal frequency is 2.4MHz, with 53% bandwidth and 170° phase.
The acoustic focusing depth is 32mm for the SV45° mode, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 32mm depth inspected in SV45° mode.

From the 5 previous figures, there is a good agreement between the results from the measurements and the results from CIVA software. It appears that the amplitude of the main echo is well estimated with less than 2dB in the worst case for nearly all the studied configurations.

Just one case shows more discrepancy, it correponds to the SV26° mode associated with the P60° mode, which shows a discrepancy up to 4dB in near field (less than 4dB difference). It is due to a splitting in the SV beam at the water/steel interface, which is not observed experimentally. The SV beam splitting is due to limitations of the model. The strong variation of the transmission coefficient close to the critical angle splits the SV incident beam. Some waves are not considered, which reduces the amplitude of the SV simulated echo.

Split of the SV beam for a Ø12.7mm circular contact probe generating P45° waves.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Mono-element immersion probe 4.5MHz, Ø12.7mm

For the Ø12.7mm circular immersion probe at 4.5MHz, the SV45° to SV60° waves are used for inspection. The input signal frequency is 4.5MHz, with 73% bandwidth and 270° phase.

The acoustic focusing depth is 26mm for the SV45° mode, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 32mm depth inspected in SV45° mode.

There is a good agreement between the results from the measurements and the results from CIVA software. It can just be noticed that CIVA underestimates the echo in the very near field of around 2dB.

Mono-element immersion probe 4.7MHz, Ø6.35mm

For the Ø6.35mm circular immersion probe at 4.7MHz, different waves are used for inspection (P and SV modes, from 45° to 60°). The water path is 25mm for all cases except one where it is 10mm. The input signal frequency is 4.7MHz, with 56% bandwidth and 255° phase.
The acoustic simulated beam is illustrated below for the P45° mode with 25mm waterpath.

The results are calibrated versus the Ø2mm SDH at 4mm depth inspected in P45° mode with 25mm water path.

Whatever the incidence angle and the water path, the simulated results fit measurements with less than 2dB discrepancy.
It can just be noticed with this crystal size that for the SV45° mode, the discrepancy tends to increase at larger depths, as it was already observed with the same crystal size at 2.25Mhz

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Conclusion

Results show a good agreement with generally less than 2dB difference for mono-element immersion probes.
The discrepancy may be a little higher in the near field. This is due limitations of the model and is explained in details for rectangular contact probes.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Side Drilled Holes at different depths and Contact probes

Global overview:

Configuration

Comparisons between measured and simulated data are also done with contact probes on the steel mock-up with Ø2mm SDH.
The measurements are performed upon a planar steel block containing Ø2mm SDH from 4 to 60mm depth with 4mm steps. As a reminder, the steel parameters are: density 7.9, P waves velocity: 5900m/s and SV waves velocity: 3230m/s. Since the SDH are inspected perpendicularly to their axis, the SOV interaction model is considered.


The following picture presents the mock-up that is used.

Different contact probes are used in Pulse Echo mode:

The results corresponding to each probe are available by clicking on the size of the probe.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Results

For each probe, the simulated P or/and SV beam radiated in the specimen and in the incidence plane is displayed for at least one configuration. For each probe, the calibration defect is the closest SDH to the focusing depth.
The superimposition of measured and simulated curves of maximal relative amplitude of the P and SV specular echoes of the SDH versus the SDH depths are presented as the next figures.

Mono-element contact probe 2.0MHz, 20*22mm, SV45°

For the 20*22mm rectangular SV45° contact probe at 2MHz, the SV45° mode is used for inspection. The input signal frequency is 2.0MHz, with 41% bandwidth and 75° phase.
The acoustic focusing depth is 53mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 52mm depth.

Mono-element contact probe 2.0MHz, 20*22mm, SV60°

For the 20*22mm rectangular SV45° contact probe at 2MHz, the SV60° mode is used for inspection. The input signal frequency is 2.0MHz, with 40% bandwidth and 0° phase.
The acoustic focusing depth is 32mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 32mm depth.

For both probes, the curves show a good agreement between simulated and measured results in the far field zone and less than 2dB difference from the focal depth to the half of the focal depth.


The discrepancy is higher in the very near field.

Click here for more information.

Discrepancies greater than 2dB occur in the very near field (up to 50% of the focal depth). In order to understand the behaviour of the defect responses for the smaller depths, the field radiated in the specimen has been computed in several zones perpendicular to the SV axis and along this axis.






These figures show the interferences between the planar wave and the edge waves of the probe leading to interferences patterns observed in the computation zones at the lower depths (within the near field distance of the probe). These interferences give rise to a lower amplitude of the field. Therefore the measured responses of the flaws seem to be coherent with this field computation. However the defect response module is based on a simplification of the radiated field. This simplification is less valid in the near field so that this additional phenomenon may cause this discrepancy with the model.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Mono-element contact probe 2.25MHz, Ø12.7mm

For the Ø12.7mm circular contact probe at 2.25MHz, the P45°, P60° and SV45° waves are used for inspection. The input signal frequency is 2.25MHz, with 50% bandwidth and 280° phase for the P waves. The input signal is the inverse P45° experimental direct specular echo of a Ø3mm FBH, tilted at 45° and located at 30mm depth for the SV45° waves. P beams are associated with lower incidence SV beams: SV22° for P45° and SV26° for P60°.
The acoustic focusing depths are 8mm, 26mm, 3mm, 15mm and 17mm respectively for the P45°, SV22°, P60°, SV26° and SV45° modes, deduced from the simulated beam as illustrated below.

	P45°		P60°
	SV22°		SV26°
SV45°

The results are calibrated versus the Ø2mm SDH at 8mm, 4mm and 20mm depth for respectively P45°, P60° and SV45° waves.

The amplitudes of the echoes from P waves are well predicted, the maximum discrepancy being less than 2dB. The amplitudes of the echoes from SV waves show higher discrepancies in the near field zone (less than 5dB) but are often correctly estimated.

Mono-element contact probe 2.25MHz, Ø6.35mm

For the Ø6.35mm circular contact probe at 2.25MHz, the SV45° mode is used for inspection. The input signal frequency is 2.25MHz, with 44% bandwidth and 147° phase.

The acoustic focusing depth is 3mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 4mm depth.

The discrepancy between simulated and measured results increases with the SDH depths but always less than 3dB, there is a good agreement.

Mono-element contact probe 4.35MHz, Ø12.7mm

For the Ø12.7mm circular contact probe at 4.35MHz, the SV45° mode is used for inspection. The input signal frequency is 4.3MHz, with 71% bandwidth and 330° phase.

The acoustic focusing depth is 31mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 36mm depth.

It can be observed that the simulated amplitudes are smoother than the experimental amplitudes which highlights probably a slight discrepancy in the measurements. There is an overall good agreement, the maximum discrepancy is less than 2dB, and from 20mm to 60mm depth, the discrepancy is always less than 1dB.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Mono-element contact probe 4.8MHz, Ø6.35mm

For the Ø6.35mm circular contact probe at 4.8MHz, the SV45° mode is used for inspection. The input signal frequency is 4.8MHz, with 45% bandwidth and 90° phase.
The acoustic focusing depth is 8mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 8mm depth.

The discrepancy is always less than 2dB, there is an overall good agreement.

Mono-element contact probe 5.0MHz, Ø6.35mm

For the Ø6.35mm circular contact probe at 5MHz, the P45° mode is used for inspection. The input signal is the inverse P45° experimental direct specular echo of a Ø2mm SDH at 8mm depth.
The acoustic focusing depths is 3mm and 12mm for respectively the P45° and SV22° modes, deduced from the simulated beam as illustrated below.

	P45°		SV22°

The results are calibrated versus the Ø2mm SDH at 4mm depth.

The discrepancy is less than 2dB for the 45° beam in the far field, there is an overall good agreement. In the near field zone, the SV beam associated with the P45° beam shows a discrepancy between 2 and 4dB, the simulation underestimates the echo.

At 32mm and 36mm depth, both SV22° and P45° experimental echoes are a little under-estimated since the SDH are damaged by rust. This block is no longer used for experimental validation.

Conclusion

Results shows a good agreement with generally less than 2dB difference in this case of SDH and mono-element contact probes. For a given probe, it can also be concluded that an SDH echo obtained for a given mode, can be used as calibration echo for all the other modes generated by the same probe with different settings.


The hypothesis of non distorted waveform on flaw, considered by Civa models, is generally valid in the far field zone or the focal zone but less in the near field zone, which leads to some discrepancies (less than 5dB) when considering flaws in the near or the very near field (from 0 to 50% of the focusing depth).



Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Side Drilled Holes at different depths and Phased-Array probes

Comparisons between experimental and simulated data are also done with phased-array contact probes on the steel mock-up with Ø2mm SDH.


Global overview:

Configuration

This validation experiment also deals with Ø 2mm SDH at different depths. The measurements are performed upon the planar steel block containing Ø2mm SDH from 4 to 60mm depth with 4mm steps. As a reminder the steel parameters are: density 7.9, P waves velocity: 5900m/s and SV waves velocity: 3230m/s. Since the SDH are inspected perpendicularly to their axis, the SOV interaction model is considered.


The following picture presents the mock-up that is used.

The 2 phased array probes that are used in Pulse Echo mode are:

The results corresponding to each probe are available by clicking on the number of elements of the probe.

Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

Results

Phased Array contact probe of 28 elements with 0.7mm pitch at 5MHz

For the phased array contact probe at 5MHz, different focal laws are used with SV waves for inspection with 28 active elements out of 48. The first experiment uses a focusing at 20mm depth and a beam angle from 45° to 60°. The second one uses a beam steering of 45° and a focusing from 4 to 24mm depth.
The acoustic focusing depth is 20mm for the SV45° beam steering, deduced from the simulated beam as illustrated below.

The results from the first experiment show the influence of the beam orientation.
The results are calibrated versus the Ø2mm SDH at 20mm depth for SV45° waves focused at 20mm.

For all the angles, the amplitudes of the echoes are well estimated with less than 2dB discrepancies.

The results from the second experiment show the influence of the focusing depth.
The results are calibrated versus the Ø2mm SDH at 20mm depth for SV45° waves focused at 20mm.

With delay laws focusing at different depths, there is a good agreement. The maximum discrepancy is 3dB for SDH at 8mm depth and focusing at 20mm depth, and less than 2dB in all the other cases.

Phased Array contact probe of 20 elements with 0.7mm pitch at 5MHz

The phased-array probe is used to generate a P45° beam at 5MHz with 20 active elements out of 48. The acoustic focusing depth is 20mm, deduced from the simulated beam as illustrated below.

The results are calibrated versus the Ø2mm SDH at 20mm depth.

With less than 1dB discrepancy, there is a very good agreement between experiment and simulation.

Conclusion

Results shows a very good agreement with less than 2dB difference in the cases of SDH and contact phased-array probes.


Back to the top
Back to the UT Module Validation menu
Back to the Validation menu

The CIVA UT module allows to compute ultrasonic field propagation in a given component and to simulate defect response, for a wide range of transducers (single element or phased-array). Results are displayed in conventional imaging tools (Ascan, Bscan, Cscan…) associated to ray tracing and advanced reconstruction tools allowing to compare simulation with acquisition data.

In the framework of validation, experiments have been performed in order to evaluate the reliability and accuracy of CIVA predictions in the « UT Defect Response » module. CIVA results are often compared to experimental results but also sometimes they might be compared to results from other models. The reliability and accuracy are estimated by comparing amplitudes of the echoes. Those amplitudes are given relatively to the amplitude of a specular echo obtained on a calibration reflector (Side Drilled Hole (SDH), Flat Bottom Hole (FBH)…). You can find more information about the calibration method here.

The first part of the validation work deals with those reference reflectors in the aim of defining which ones can be used with confidence for various probes.

The second part concerns « classical » echoes in NDE: corner echoes from backwall or surface breaking notches in planar specimens.

The validation cases presented here deals with quite simple cases (canonical component, isotropic material…). A second validation campaign has begun at the CEA. These validations will deal with more complex inspections in terms of part geometry, type of inspections…

NB: in those validation pages, Transversal waves are called SV (Shear Vertical) waves, and Longitudinal waves are called P (Pressure) waves.

Achieved

The following experiments have been led for validation purposes.

Echoes obtained from calibration reflectors:

• Ø2mm Side Drilled Holes at different depths
• Flat Bottom Holes of different diameters and at different depths
• Comparison of Side Drilled Holes with Flat Bottom Holes

Corner echoes:

• Corner echoes obtained on rectangular notches in planar specimens

In progress

• TOFD
• Inspection with Phased Array probes
• Inspection with Focused probes
• Flaws in cylindrical specimen
• Geometrical echoes

Complementary information

Results from different simulations can be compared according to some rules.

As Civa relies on semi-analytical models, many approximations exist and the limits must be known by the user.


Back to the Validation menu

• UT
  • QNDE Benchmark 2005
  • QNDE Benchmark 2006
  • QNDE Benchmark 2007
  • QNDE Benchmark 2008
  • QNDE Benchmark 2009
  • QNDE Benchmark 2010
• ET
  • QNDE Benchmark 2007
  • QNDE Benchmark 2008

Par ailleurs, Les modèles et performances de CIVA sont testés par une séquence de validation constituée par plus de 1000 tests effectués au Laboratoire de Test et Validation du CEA.