Ultrasonic Testing - COMSOL.com

Excerpt from the Proceedings of the 2012 COMSOL Conference in MilanCalibration of Ultrasonic Testingfor faults detection in stone masonryM. Usai 1 , S. Carcangiu 1 , and G. Concu 21Department of Electric and Electronic Engineering2Department of Civil Engineering, Environmental and ArchitectureUniversity of Cagliari, Sardinia, ItalyCOMSOL CONFERENCE EUROPE 2012

• IntroductionNon-Destructive TestingUltrasonic Testing• Experimental sessionOutline• Simulations with COMSOL Multiphisycs• Comparison between experimental and simulated data• Results and conclusionsCOMSOL CONFERENCE EUROPE 2012

IntroductionThe Research Group in Electrical Engineering of Cagliari, withcolleagues of the Department of Civil Engineering, hasdeveloped skills and knowledge in the field of signalprocessing in Non-Destructive-Testing-Techniques NDTs onhomogeneous materials (metals) and non-homogeneousmaterials (building materials).COMSOL CONFERENCE EUROPE 2012

IntroductionIn the field of assessment methodologies, particularimportance is given to Non-Destructive-Testing-Techniques NDTs,which aspire to achieve the highest number of information aboutmaterials and structures without altering their condition.The implementation of algorithms for signals post-processingconsists in extracting information from test signals derived fromthe non-destructive tests.COMSOL CONFERENCE EUROPE 2012

Non-Destructive TestingAmong NDTs, Ultrasonic Testing exploits the transmission andreflection characteristics of mechanical waves with appropriatefrequencies passing through the investigated item.EMITTER TRANSDUCERWALLPULSERRECEIVERRECEIVER TRANSDUCERDIGITALOSCILLOSCOPECOMSOL CONFERENCE EUROPE 2012

Ultrasonic TestingElastic waves propagate in different manner through solidmaterials and cavities, thus enabling fault detection.For example, the figure shows how in the presence of a fault,the transmitted wave gives rise to a signal of lower amplitude.COMSOL CONFERENCE EUROPE 2012

Ultrasonic TestingThe testing method is the Direct Transmission Technique (EN14579 2004).In this method, the ultrasonic wave is transmitted by a transducer(Transmitter) through the test object and received by a secondtransducer (Receiver) on the opposite side of the structure.COMSOL CONFERENCE EUROPE 2012

Ultrasonic TestingDue to media dissipative effect, elastic waves are strongly attenuatedthe emission cone of the signal source can be less divergent as possible.Acoustic field of the transducerThe sound field of a transducer is divided into two zones: the near field and the far field.Near field: region directly in front of the transducer where the amplitudegoes through a series of maxima and minima and ends at the last maximum,at distance N from the transducer.Far field: area beyond N, where the sound field pressuregradually drops to zero.Because of the variations within the near fieldit can be difficult to accurately evaluate flawsusing amplitude based techniques.D diameter of the Transducer sourceλ wavelength of the waveN 2D4λ Near field distanceCOMSOL CONFERENCE EUROPE 2012

Ultrasonic TestingDue to media dissipative effect, elastic waves are strongly attenuatedthe emission cone of the signal source can be less divergent as possible.Beam Spread and Half AngleAll ultrasonic beams divergethe beam spread is a function of wavelength1.2sin DIf D >> λ the wave is emitted with a not very divergent cone.As λ is in inverse proportion to frequency f, it is understandable howhigh frequency signals enable waves to be highly directional.COMSOL CONFERENCE EUROPE 2012

Ultrasonic TestingThese studies need the highest number of signals as possible.Laboratory tests must be accurate and repeated several times toobtain reliable signals.↓In this phase of the work becomes important to simulateeffectively the real models with a suitable FEM code.COMSOL CONFERENCE EUROPE 2012

Experimental sessionFront view of the wallHorizontal plane sectionVertical plane sectionThe Ultrasonic Techniquehas been carried out on atrachyte stone masonrywith a cavity inside.COMSOL CONFERENCE EUROPE 2012

Experimental sessionThe wall is 90 cm wide, 62 cmhigh and 38 cm thick, and it ismade of trachyte blocks sized20 × 38 × 12 cm 3 joined withmortar.To map the faults308 points for the emittertransducer and 308 pointsfor the receiver transducerhave been arranged in a gridof 14 x 22 nodes in theopposite surfaces of the wall.COMSOL CONFERENCE EUROPE 2012

Experimental set-upUltrasonic measurements have been carried out using theultrasonic test equipment Pundit Lab+, developed byProceq, with standard 54 KHz transducers .COMSOL CONFERENCE EUROPE 2012

Voltage [V]Experimental sessionThe energizing signal is a square wave with high input voltage of 500 V,which develops in a time interval of 9.3μs.The high intensity of this signal allowed signals on the other side of the wall to bedetected even if strongly attenuated.COMSOL CONFERENCE EUROPE 2012

Experimental sessionA signal has been acquired in each point of the grid of receivers. From each node ofthe grid, several features in both time and frequency domain can been extracted.Time domain• Propagation velocity• Maximum peak-peak amplitude• Maximum absolute amplitude• Minimum absolute amplitude• Mean value of normalized amplitude• Variance of normalized amplitude• Time–reversibility• Third order auto-covariance• Mean value of normalized signal envelope• Variance of normalized signal envelope• Local rise time from 25% level to peak ofnormalized signal envelope• Local rise time from 50% level to peak ofnormalized signal envelope• Signal power• Travel timeFrequency Domain• Spectrum principal frequency• Spectrum normalized maximum amplitude• Input signal attenuation• Spectrum central frequency• Spectrum Bandwidth• Spectrum normalized mean value• Spectrum normalized varianceSome of the features that are normally considered to extract information about faults from the signalCOMSOL CONFERENCE EUROPE 2012

Experimental sessionReceived signals have different patterns depending on the nature of the materials crossed.It is possible to distinguish different typesof signal paths, corresponding to differentmaterials regions in the wall :trachyte path (pale blue) 2trachyte-mortar path (yellow) 1 and 3trachyte-mortar-air path (brown) 4Map associated to the different materialswhich make up the wall.COMSOL CONFERENCE EUROPE 2012

Experimental signals crossing the 4 paths (Time domain)To different pathscorrespond differentsignals with unlikeamplitudesandharmonic content.12Due to edge effects,signals travellingclose to the boundaryof a material regionare different fromthose crossing theinner part of thesame region.3 4COMSOL CONFERENCE EUROPE 2012

Spectra(Frequencydomain)The harmoniccontent of thesignals takes placearound theresonant frequencyof thetransducer (54 KHz)and differs for eachpath.1324COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsThe numerical analyses have performed using a Pressure Acoustics model and thePiezoelectric Devices Interface belonging to the COMSOL Acoustics Module.In order to decrease the size of the resolution matrices, with significant reduction ofprocessing time and computational errors :the frequency domain rather than the time domain has been chosen to solve themodel. The sound field is described and solved by the pressure p, so we cansimulate only one transducer (the emitter)two sub-models coupled using the Extrusion Model Couplings option of COMSOLhave been considered:one for the emission transducer (2D axisymmetric) and another for the wall (3D).COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsThe wall and the emitter transducer have been simulated with two submodelsCOMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsTwo sub-models coupledWall model (3D)Emission transducer model(2D axisymmetric)COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsThe sound field is described and solved by the pressure p.The acoustic pressure in the time domain p(t) is obtained as a function of the spectralcomponents as:pfNt p f cos2f t p f i fi1minampfmaxiphaseiN number of components of the spectra in the frequency rangeCOMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsAnalytic model of Pressure Acoustics, Frequency Domain interfaceThe acoustic pressure p(t ) in a medium is governed by the following equationthat is an inhomogeneous Helmholtz equation:1 pkteqcpptp2 c b; k• ρ is the density of the material [kg/m 3 ],• c is the speed of sound [m/s];• Q [1/s 2 ] is a monopole acoustic sourcek2eqptc c Qj; ccc2c2COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsAnalytic model of Piezoelectric Devices interfaceThe mathematical formulation of this model is described by the equations:2 u D VF evjwhere• F v is deformation gradient,• ρ V is the volume charge density,• u is the displacement,• D is the electric displacement and• σ is the stress-charge.COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsAt the interface between the transducer and the wall, the boundarycondition for the acoustics interface is that the pressure is equal to thenormal acceleration of the solid domain: 1n 0p anwhere a n is the normal acceleration.This drives the pressure in the wall domain.COMSOL CONFERENCE EUROPE 2012

Simulations with COMSOL MultiphisycsThe implementation of the Comsol Model hasrequired a series of simulations to develop all theparameters, until than the results wereconsistent with the experimental ones.COMSOL CONFERENCE EUROPE 2012

Simulated signals crossing the 4 paths (Time domain)Simulated signalsshow aperformance ingood agreementwith experimentalsignals12Amplitudes in thetime domain arescaled to theaverage gain of theemissiontransducer3 4COMSOL CONFERENCE EUROPE 2012

Simulated Spectra (Frequency domain)Spectral bandwidthare shifted tohigher values , asexpected1234COMSOL CONFERENCE EUROPE 2012

ResultsSimilarly to what is found for the measured signals:also the simulated signals present different performances for thedifferent types of paths both in time and frequency domain, andthis allows to post-process them to map the macro defects.COMSOL CONFERENCE EUROPE 2012

ConclusionsThe purpose of this work was to create a model of a prototype of a trachyte wall, builtand tested in the laboratory, in order to simulate the ultrasonic signal propagation andthus be able to perform parametric studies freed from the experimental sessions.Model calibration has been achieved by comparison of a significant sample ofsimulated signal to the correspondent signals obtained from the experimental sessionsResults show that the implemented COMSOL model is suitable to effectively simulatethe ultrasonic signals transmission through the stone wallThe simulated signals can be used to obtain, through post-processing analysis, mapsfor detecting the presence of macro defects with cluster algorithms or non supervisedNeural Networks as Self organizing Maps (SOM)COMSOL CONFERENCE EUROPE 2012

ConclusionsThe purpose of this work was to create a model of a prototype of a trachyte wall, builtand tested in the laboratory, in order to simulate the ultrasonic signal propagation andthus be able to perform parametric studies freed from the experimental sessions.Model calibration has been achieved by comparison of a significant sample ofsimulated signal to the correspondent signals obtained from the experimental sessionsResults show that the implemented COMSOL model is suitable to effectively simulatethe ultrasonic signals transmission through the stone wallThe simulated signals can be used to obtain, through post-processing analysis, mapsfor detecting the presence of macro defects with cluster algorithms or non supervisedNeural Networks as Self organizing Maps (SOM)COMSOL CONFERENCE EUROPE 2012

Thanks for your kind attentionFor further information:musai@diee.unica.its.carcangiu@diee.unica.itCOMSOL CONFERENCE EUROPE 2012

Thanks for your kind attentionThanks for your kind attentionCagliariCOMSOL CONFERENCE EUROPE 2012