SAM
https://sam.ensam.eu:443
The DSpace digital repository system captures, stores, indexes, preserves, and distributes digital research material.Tue, 20 Feb 2024 18:34:06 GMT2024-02-20T18:34:06ZCharacterization of identification errors and uses in localization of poor modal correlation
http://hdl.handle.net/10985/11675
Characterization of identification errors and uses in localization of poor modal correlation
MARTIN, Guillaume; BALMES, Etienne; CHANCELIER, Thierry
While modal identification is a mature subject, very few studies address the characterization of errors associated with components of a mode shape. This is particularly important in test/analysis correlation procedures, where the Modal Assurance Criterion is used to pair modes and to localize at which sensors discrepancies occur. Poor correlation is usually attributed to modeling errors, but clearly identification errors also occur. In particular with 3D Scanning Laser Doppler Vibrometer measurement, many transfer functions are measured. As a result individual validation of each measurement cannot be performed manually in a reasonable time frame and a notable fraction of measurements is expected to be fairly noisy leading to poor identification of the associated mode shape components. The paper first addresses measurements and introduces multiple criteria. The error measures the difference between test and synthesized transfer functions around each resonance and can be used to localize poorly identified modal components. For intermediate error values, diagnostic of the origin of the error is needed. The level evaluates the transfer function amplitude in the vicinity of a given mode and can be used to eliminate sensors with low responses. A Noise Over Signal indicator, product of error and level, is then shown to be relevant to detect poorly excited modes and errors due to modal property shifts between test batches. Finally, a contribution is introduced to evaluate the visibility of a mode in each transfer. Using tests on a drum brake component, these indicators are shown to provide relevant insight into the quality of measurements. In a second part, test/analysis correlation is addressed with a focus on the localization of sources of poor mode shape correlation. The MACCo algorithm, which sorts sensors by the impact of their removal on a MAC computation, is shown to be particularly relevant. Combined with the error it avoids keeping erroneous modal components. Applied after removal of poor modal components, it provides spatial maps of poor correlation, which help localizing mode shape correlation errors and thus prepare the selection of model changes in updating procedures.
Sun, 01 Jan 2017 00:00:00 GMThttp://hdl.handle.net/10985/116752017-01-01T00:00:00ZMARTIN, GuillaumeBALMES, EtienneCHANCELIER, ThierryWhile modal identification is a mature subject, very few studies address the characterization of errors associated with components of a mode shape. This is particularly important in test/analysis correlation procedures, where the Modal Assurance Criterion is used to pair modes and to localize at which sensors discrepancies occur. Poor correlation is usually attributed to modeling errors, but clearly identification errors also occur. In particular with 3D Scanning Laser Doppler Vibrometer measurement, many transfer functions are measured. As a result individual validation of each measurement cannot be performed manually in a reasonable time frame and a notable fraction of measurements is expected to be fairly noisy leading to poor identification of the associated mode shape components. The paper first addresses measurements and introduces multiple criteria. The error measures the difference between test and synthesized transfer functions around each resonance and can be used to localize poorly identified modal components. For intermediate error values, diagnostic of the origin of the error is needed. The level evaluates the transfer function amplitude in the vicinity of a given mode and can be used to eliminate sensors with low responses. A Noise Over Signal indicator, product of error and level, is then shown to be relevant to detect poorly excited modes and errors due to modal property shifts between test batches. Finally, a contribution is introduced to evaluate the visibility of a mode in each transfer. Using tests on a drum brake component, these indicators are shown to provide relevant insight into the quality of measurements. In a second part, test/analysis correlation is addressed with a focus on the localization of sources of poor mode shape correlation. The MACCo algorithm, which sorts sensors by the impact of their removal on a MAC computation, is shown to be particularly relevant. Combined with the error it avoids keeping erroneous modal components. Applied after removal of poor modal components, it provides spatial maps of poor correlation, which help localizing mode shape correlation errors and thus prepare the selection of model changes in updating procedures.Review of model updating processes used for brake components
http://hdl.handle.net/10985/10921
Review of model updating processes used for brake components
MARTIN, Guillaume; BALMES, Etienne; VERMOT DES ROCHES, Guillaume; CHANCELIER, Thierry
To be confident in the prediction capability of a model, verification and validation steps are classically performed. Verification checks that the model is properly solved. Since the model used are fairly standard, this is not issue for brake components. Validation checks the relation between model and experiments on actual structures. Here geometry measurements and vibration tests are considered. The study seeks to perform a systematic review of how test quality is evaluated, and models are correlated and then updated. This will give a solid basis to define clear and easily used validations protocols for brake components where prediction of modes and their stability in the manufacturing process is often deemed critical. Updating the geometry before updating the material properties is shown to be very important: the residual error on frequencies is smaller and no bias is introduced in the estimated material properties. Proper pairing of modeshapes is important for broadband comparisons and the MAC criterion is used. Intermediate steps: experimental topology correlation using easy tools with accuracy evaluation, estimation of errors on test shapes, handling of mode crossing, are sources of errors that are analyzed. For the updating of contact properties, where many parameters may need update, the use of model reduction is shown to allow a major speed-up of parametric studies.
Thu, 01 Jan 2015 00:00:00 GMThttp://hdl.handle.net/10985/109212015-01-01T00:00:00ZMARTIN, GuillaumeBALMES, EtienneVERMOT DES ROCHES, GuillaumeCHANCELIER, ThierryTo be confident in the prediction capability of a model, verification and validation steps are classically performed. Verification checks that the model is properly solved. Since the model used are fairly standard, this is not issue for brake components. Validation checks the relation between model and experiments on actual structures. Here geometry measurements and vibration tests are considered. The study seeks to perform a systematic review of how test quality is evaluated, and models are correlated and then updated. This will give a solid basis to define clear and easily used validations protocols for brake components where prediction of modes and their stability in the manufacturing process is often deemed critical. Updating the geometry before updating the material properties is shown to be very important: the residual error on frequencies is smaller and no bias is introduced in the estimated material properties. Proper pairing of modeshapes is important for broadband comparisons and the MAC criterion is used. Intermediate steps: experimental topology correlation using easy tools with accuracy evaluation, estimation of errors on test shapes, handling of mode crossing, are sources of errors that are analyzed. For the updating of contact properties, where many parameters may need update, the use of model reduction is shown to allow a major speed-up of parametric studies.Squeal measurement with 3D Scanning Laser Doppler Vibrometer: handling of the time varying system behavior and analysis improvement using FEM expansion
http://hdl.handle.net/10985/13856
Squeal measurement with 3D Scanning Laser Doppler Vibrometer: handling of the time varying system behavior and analysis improvement using FEM expansion
MARTIN, Guillaume; CHANCELIER, Thierry; VERMOT DES ROCHES, Guillaume; BALMES, Etienne
In the presence of squeal, Operational Deflection Shapes (ODS) are classically performed to analyze behavior. A simple numeric example is used to show that two real shapes should dominate the response. This justifies an ad-hoc procedure to extract main shapes from the real brake time measurements. The presence of two shapes is confirmed despite variations with wheel position and reproducibility tests. To obtain a high spatial density measurement, 3D Scanning Laser Doppler Vibrometer is interesting but leads to iterative measurements on a time-varying system. An algorithm to merge sequential measurement and extract main shapes is detailed. Even with a high-density 3D SLDV measurement, shapes characterizing the squeal event are still only known on accessible surfaces. Minimum Dynamic Residual Expansion (MDRE) is thus finally used to estimate motion on a full FE mesh which eases interpretation and highlights areas where the test and the model contain errors.
Mon, 01 Jan 2018 00:00:00 GMThttp://hdl.handle.net/10985/138562018-01-01T00:00:00ZMARTIN, GuillaumeCHANCELIER, ThierryVERMOT DES ROCHES, GuillaumeBALMES, EtienneIn the presence of squeal, Operational Deflection Shapes (ODS) are classically performed to analyze behavior. A simple numeric example is used to show that two real shapes should dominate the response. This justifies an ad-hoc procedure to extract main shapes from the real brake time measurements. The presence of two shapes is confirmed despite variations with wheel position and reproducibility tests. To obtain a high spatial density measurement, 3D Scanning Laser Doppler Vibrometer is interesting but leads to iterative measurements on a time-varying system. An algorithm to merge sequential measurement and extract main shapes is detailed. Even with a high-density 3D SLDV measurement, shapes characterizing the squeal event are still only known on accessible surfaces. Minimum Dynamic Residual Expansion (MDRE) is thus finally used to estimate motion on a full FE mesh which eases interpretation and highlights areas where the test and the model contain errors.Updating and design sensitivity processes applied to drum brake squeal analysis
http://hdl.handle.net/10985/10922
Updating and design sensitivity processes applied to drum brake squeal analysis
MARTIN, Guillaume; BALMES, Etienne; VERMOT DES ROCHES, Guillaume; CHANCELIER, Thierry
Squeal occurrences are quite common in brakes in production and involve coupling of modes. Detailed understanding of vibration patterns typically requires FEM models updated using test results. The process used at Chassis Brakes International typically starts by updating components so that the main sources of variability are associated with junctions. A modeling strategy allowing the practical analysis of the impact of junctions is proposed and illustrated on the case of a drum brake assembly. As the level of uncertainty/design freedom is fairly large for junctions, the evolution of modal properties is difficult to interpret. The notion of component modes within a rigid assembly is thus introduced and shown to provide an appropriate interpretation of changes in a system with multiple modal crossings. The analysis of possible forced responses is finally shown to lead to relevant interpretation of possibly interesting designs or problematic instances of a variable component.
Fri, 01 Jan 2016 00:00:00 GMThttp://hdl.handle.net/10985/109222016-01-01T00:00:00ZMARTIN, GuillaumeBALMES, EtienneVERMOT DES ROCHES, GuillaumeCHANCELIER, ThierrySqueal occurrences are quite common in brakes in production and involve coupling of modes. Detailed understanding of vibration patterns typically requires FEM models updated using test results. The process used at Chassis Brakes International typically starts by updating components so that the main sources of variability are associated with junctions. A modeling strategy allowing the practical analysis of the impact of junctions is proposed and illustrated on the case of a drum brake assembly. As the level of uncertainty/design freedom is fairly large for junctions, the evolution of modal properties is difficult to interpret. The notion of component modes within a rigid assembly is thus introduced and shown to provide an appropriate interpretation of changes in a system with multiple modal crossings. The analysis of possible forced responses is finally shown to lead to relevant interpretation of possibly interesting designs or problematic instances of a variable component.Improved Modal Assurance Criterion using a quantiﬁcation of identiﬁcation errors per mode/sensor
http://hdl.handle.net/10985/8592
Improved Modal Assurance Criterion using a quantiﬁcation of identiﬁcation errors per mode/sensor
MARTIN, Guillaume; BALMES, Etienne; CHANCELIER, Thierry
Scanning laser vibrometer measurements generate detailed maps of modal characteristics in normal or 3D directions. Since many points are measured, individual validation of cannot be performed manually in a reasonable time frame and a notable fraction of measurements is expected to be fairly noisy. The key new notion of the paper is a quantiﬁcation of identiﬁcation error and level of contribution for each mode and each sensor based on the comparison of measured and synthesized transfers around each resonance. These criteria are shown to allow efﬁcient analysis of the validity of large measurement sets to provide an automated procedure to select sensors that should be kept for each mode. This quantiﬁcation, being performed before correlation, provides a priori estimates of sources of poor correlation associated with the identiﬁcation process. It thus becomes possible to provide improved Modal Assurance Criterion estimations where, for each modeshape, sensors known to be incorrectly identiﬁed can be discarded. The 3D vibrometer scan of a brake component is used to illustrate the proposed strategies.
Wed, 01 Jan 2014 00:00:00 GMThttp://hdl.handle.net/10985/85922014-01-01T00:00:00ZMARTIN, GuillaumeBALMES, EtienneCHANCELIER, ThierryScanning laser vibrometer measurements generate detailed maps of modal characteristics in normal or 3D directions. Since many points are measured, individual validation of cannot be performed manually in a reasonable time frame and a notable fraction of measurements is expected to be fairly noisy. The key new notion of the paper is a quantiﬁcation of identiﬁcation error and level of contribution for each mode and each sensor based on the comparison of measured and synthesized transfers around each resonance. These criteria are shown to allow efﬁcient analysis of the validity of large measurement sets to provide an automated procedure to select sensors that should be kept for each mode. This quantiﬁcation, being performed before correlation, provides a priori estimates of sources of poor correlation associated with the identiﬁcation process. It thus becomes possible to provide improved Modal Assurance Criterion estimations where, for each modeshape, sensors known to be incorrectly identiﬁed can be discarded. The 3D vibrometer scan of a brake component is used to illustrate the proposed strategies.MDRE: an efficient expansion tool to perform model updating from squeal measurements
http://hdl.handle.net/10985/16935
MDRE: an efficient expansion tool to perform model updating from squeal measurements
MARTIN, Guillaume; VERMOT DES ROCHES, Guillaume; BALMES, Etienne; CHANCELIER, Thierry
In brake FEM, model updating is often needed to improve the model accuracy and well describe problematic phenomena such as the squeal. To avoid performing a full model updating which is often time consuming, the use of the Minimum Dynamic Residual Expansion method is proposed to help building the updating strategy. The procedure proposed in this paper is evaluated on a disc brake system, using experimental measurements and the nominal model as input data. From experimental squeal measurements, two shapes are extracted and expanded on the current model. The evaluation of the residual error of model shows areas where the model is wrong and guides through the definition of sensitive parameters which need to be updated. Once the model is parameterized, a model reduction strategy is proposed for further computations to be performed in a time compatible with industrial processes. A parametric study is then achieved: the expansion is computed for all the combinations of the chosen parameters. It is finally possible to navigate through the expansion results for all the parameters, evaluate the evolution of the model accuracy and extract the best combination which improves the model representability.
Tue, 01 Jan 2019 00:00:00 GMThttp://hdl.handle.net/10985/169352019-01-01T00:00:00ZMARTIN, GuillaumeVERMOT DES ROCHES, GuillaumeBALMES, EtienneCHANCELIER, ThierryIn brake FEM, model updating is often needed to improve the model accuracy and well describe problematic phenomena such as the squeal. To avoid performing a full model updating which is often time consuming, the use of the Minimum Dynamic Residual Expansion method is proposed to help building the updating strategy. The procedure proposed in this paper is evaluated on a disc brake system, using experimental measurements and the nominal model as input data. From experimental squeal measurements, two shapes are extracted and expanded on the current model. The evaluation of the residual error of model shows areas where the model is wrong and guides through the definition of sensitive parameters which need to be updated. Once the model is parameterized, a model reduction strategy is proposed for further computations to be performed in a time compatible with industrial processes. A parametric study is then achieved: the expansion is computed for all the combinations of the chosen parameters. It is finally possible to navigate through the expansion results for all the parameters, evaluate the evolution of the model accuracy and extract the best combination which improves the model representability.A Structural Dynamics Modification Strategy based on Expanded Squeal Operational Deflection Shapes
http://hdl.handle.net/10985/23306
A Structural Dynamics Modification Strategy based on Expanded Squeal Operational Deflection Shapes
MARTIN, Guillaume; BALMES, Etienne; CHANCELIER, Thierry; THOUVIOT, Sylvain; LEMAIRE, Rémi
To analyze brake squeal, measurements are performed to extract Operational Deflection Shapes (ODS) characteristic of the limit cycle. The advantage of this strategy is that the real system behavior is captured, but measurements suffer from a low spatial distribution and hidden surfaces, so that interpretation is sometimes difficult. It is even more difficult to propose system modifications from test alone. Historical Structural Dynamics Modification (SDM) techniques need mass normalized shapes which is not available from an ODS measurement. Furthermore, it is very difficult to translate mass, damping or stiffness modification between sensors into physical modifications of the real system. On the model side, FEM methodology gives access to fine geometric details, continuous field over the whole system. Simple simulation of the impact of modifications is possible, one typical strategy for squeal being to avoid unstable poles. Nevertheless, to ensure accurate predictions, test/FEM correlation must be checked and model updating may be necessary despite high cost and absence of guarantee on results. To combine both strategies, expansion techniques seek to estimate the ODS on all FEM DOF using a multi-objective optimization combining test and model errors. The high number of sensors compensates for modeling errors, while allowing imperfect test. The Minimum Dynamics Residual Expansion (MDRE) method used here, ensures that the complex ODS expanded shapes are close enough to the measured motion but have smooth, physically representative, stress field, which is mandatory for further analysis. From the expanded ODS and using the model, the two underlying real shapes are mass-orthonormalized and stiffness-orthogonalized resulting in a reduced modal model with two modes defined at all model DOFs. Sensitivity analysis is then possible and the impact of thickness modifications on frequencies is estimated. This provides a novel structural modification strategy where the parameters are thickness distributions and the objective is to separate the frequencies associated with the two shapes found by expansion of the experimental ODS.
Sun, 01 May 2022 00:00:00 GMThttp://hdl.handle.net/10985/233062022-05-01T00:00:00ZMARTIN, GuillaumeBALMES, EtienneCHANCELIER, ThierryTHOUVIOT, SylvainLEMAIRE, RémiTo analyze brake squeal, measurements are performed to extract Operational Deflection Shapes (ODS) characteristic of the limit cycle. The advantage of this strategy is that the real system behavior is captured, but measurements suffer from a low spatial distribution and hidden surfaces, so that interpretation is sometimes difficult. It is even more difficult to propose system modifications from test alone. Historical Structural Dynamics Modification (SDM) techniques need mass normalized shapes which is not available from an ODS measurement. Furthermore, it is very difficult to translate mass, damping or stiffness modification between sensors into physical modifications of the real system. On the model side, FEM methodology gives access to fine geometric details, continuous field over the whole system. Simple simulation of the impact of modifications is possible, one typical strategy for squeal being to avoid unstable poles. Nevertheless, to ensure accurate predictions, test/FEM correlation must be checked and model updating may be necessary despite high cost and absence of guarantee on results. To combine both strategies, expansion techniques seek to estimate the ODS on all FEM DOF using a multi-objective optimization combining test and model errors. The high number of sensors compensates for modeling errors, while allowing imperfect test. The Minimum Dynamics Residual Expansion (MDRE) method used here, ensures that the complex ODS expanded shapes are close enough to the measured motion but have smooth, physically representative, stress field, which is mandatory for further analysis. From the expanded ODS and using the model, the two underlying real shapes are mass-orthonormalized and stiffness-orthogonalized resulting in a reduced modal model with two modes defined at all model DOFs. Sensitivity analysis is then possible and the impact of thickness modifications on frequencies is estimated. This provides a novel structural modification strategy where the parameters are thickness distributions and the objective is to separate the frequencies associated with the two shapes found by expansion of the experimental ODS.Expansion in structural dynamics : a perspective gained from success and errors in test/FEM twin building
http://hdl.handle.net/10985/23307
Expansion in structural dynamics : a perspective gained from success and errors in test/FEM twin building
BALMES, Etienne; MARTIN, Guillaume; VERMOT DES ROCHES, Guillaume; CHANCELIER, Thierry; THOUVIOT, Sylvain
Since tests only provide measurements at sensors, it is interesting to use models to estimate the response at all degree of freedom, correct measurement errors and possibly allow updating of model parameters. The paper gives an integrated perspective on methods developed by the control and structural dynamics communities and in particular methods seeking a trade-off between test and model error. The case of a measured brake squeal limit cycle is used to illustrate implementation details found to be important.
Sun, 01 May 2022 00:00:00 GMThttp://hdl.handle.net/10985/233072022-05-01T00:00:00ZBALMES, EtienneMARTIN, GuillaumeVERMOT DES ROCHES, GuillaumeCHANCELIER, ThierryTHOUVIOT, SylvainSince tests only provide measurements at sensors, it is interesting to use models to estimate the response at all degree of freedom, correct measurement errors and possibly allow updating of model parameters. The paper gives an integrated perspective on methods developed by the control and structural dynamics communities and in particular methods seeking a trade-off between test and model error. The case of a measured brake squeal limit cycle is used to illustrate implementation details found to be important.Squeal measurement using operational deflection shape. Quality assessment and analysis improvement using FEM expansion.
http://hdl.handle.net/10985/12794
Squeal measurement using operational deflection shape. Quality assessment and analysis improvement using FEM expansion.
BALMES, Etienne; VERMOT DES ROCHES, Guillaume; CHANCELIER, Thierry; MARTIN, Guillaume
In presence of squeal, Operational Deflection Shapes (ODS) are classically measured to gain understanding of brake behavior. A simple numeric example is analyzed to justify the use of time-frequency analysis and shows that two real shapes should probably dominate the response. Using measurements on a real brake, this expectation is shown to hold even in the presence of variations with wheel position as well as for reproducibility tests. For a proper relation with the model, it is desirable to also extract modes. The test campaign is used to illustrate how this can be quite difficult due to reproducibility problems. Finally, shapes characterizing the squeal event are fundamentally limited by measurable quantities. Minimum Dynamic Residual Expansion (MDRE), which estimates test motion at all FE degrees of freedom, is shown to be applicable for industrial models and gives insight of test and model imperfections.
Sun, 01 Jan 2017 00:00:00 GMThttp://hdl.handle.net/10985/127942017-01-01T00:00:00ZBALMES, EtienneVERMOT DES ROCHES, GuillaumeCHANCELIER, ThierryMARTIN, GuillaumeIn presence of squeal, Operational Deflection Shapes (ODS) are classically measured to gain understanding of brake behavior. A simple numeric example is analyzed to justify the use of time-frequency analysis and shows that two real shapes should probably dominate the response. Using measurements on a real brake, this expectation is shown to hold even in the presence of variations with wheel position as well as for reproducibility tests. For a proper relation with the model, it is desirable to also extract modes. The test campaign is used to illustrate how this can be quite difficult due to reproducibility problems. Finally, shapes characterizing the squeal event are fundamentally limited by measurable quantities. Minimum Dynamic Residual Expansion (MDRE), which estimates test motion at all FE degrees of freedom, is shown to be applicable for industrial models and gives insight of test and model imperfections.Viscoelastic homogenization of 3D woven composites with damping validation in temperature and verification of scale separation
http://hdl.handle.net/10985/20775
Viscoelastic homogenization of 3D woven composites with damping validation in temperature and verification of scale separation
CONEJOS, Florian; BALMES, Etienne; TRANQUART, Bastien; MARTIN, Guillaume; MONTEIRO, Eric
Estimation of damping can be of great importance for turbomachines, where vibration based instabilities like flutter occur. The paper discusses a numerical method to predict the homogenized viscoelastic behavior of 3D woven composites, used in fan blades, from elementary constituent behavior. Yarn and weave microstructures are considered in a two scale homogenization. The matrix and fibers are considered homogeneous with linear viscoelastic and elastic behavior respectively. Temperature and frequency dependence of matrix properties are characterized by complex moduli. Confrontation of numerical predictions with modal damping of a modified Oberst experiment, for a temperature range of -40 to 120 °C, gives good results in terms of absolute value and trends. The homogenization is formulated using matrix operations, which enables the simple use of model reduction techniques for parametric studies on temperature and leads to energy fraction analyses useful to gain understanding of how different components of the constitutive laws contribute to damping and change with temperature. Finally, since weaving patterns have a scale of a few centimeters, that is not small compared to gradients found in the experiment, exact solutions for responses to regular volume loads are used to characterize the validity of the scale separation hypothesis as a function of wavelength.
Fri, 01 Jan 2021 00:00:00 GMThttp://hdl.handle.net/10985/207752021-01-01T00:00:00ZCONEJOS, FlorianBALMES, EtienneTRANQUART, BastienMARTIN, GuillaumeMONTEIRO, EricEstimation of damping can be of great importance for turbomachines, where vibration based instabilities like flutter occur. The paper discusses a numerical method to predict the homogenized viscoelastic behavior of 3D woven composites, used in fan blades, from elementary constituent behavior. Yarn and weave microstructures are considered in a two scale homogenization. The matrix and fibers are considered homogeneous with linear viscoelastic and elastic behavior respectively. Temperature and frequency dependence of matrix properties are characterized by complex moduli. Confrontation of numerical predictions with modal damping of a modified Oberst experiment, for a temperature range of -40 to 120 °C, gives good results in terms of absolute value and trends. The homogenization is formulated using matrix operations, which enables the simple use of model reduction techniques for parametric studies on temperature and leads to energy fraction analyses useful to gain understanding of how different components of the constitutive laws contribute to damping and change with temperature. Finally, since weaving patterns have a scale of a few centimeters, that is not small compared to gradients found in the experiment, exact solutions for responses to regular volume loads are used to characterize the validity of the scale separation hypothesis as a function of wavelength.