Model validation for squeak & rattle

Model validation for squeak & rattleValidating the tailgate model on a complete vehicleApplication note

2

Polytec and Volvo Car Corporation demon-strate how suitable numerical simulation approaches make it possible to better pre-dict the acoustic performance of component connections. Besides a good methodolo-gy, the validation of the calculated results through realistic testing is crucial for model optimization.

3

Squeak and rattle

'Squeak and rattle' (S&R) refers, as the name suggests, to undesirable squeaking and rattling noises – an issue which is steadily becoming more and more important to customer satisfaction. S&R is a non-linear phenomenon which is heavily dependent on the relative displace-ments of neighboring components. A simple example of this is a rattle that only starts when two neighboring components touch each other during vibration, i.e., a gap initially present is closed by the relative displace-ment. Squeaking, on the other hand, occurs as a result of frictional movements of neighboring components which have already been in contact. The cause of this is the so-called stick-slip phenomenon which occurs when the relative displacement exceeds a specific threshold value for the material pairing.

In order to be able to detect and avoid these back-ground noises in new vehicles as early as possible, Volvo Car Corporation uses a new comprehensive S&R simulation tool (E-Line) early on in the design phase. At the crux of this simulation is an FEM calculation in the frequency domain (modal transient) whereby the rela-tive displacements along the contact/gap lines between the neighboring components are calculated and ana-lyzed. This may concern one assembly, the dashboard for example, or a complete vehicle.

One parameter which is of key importance to this calculation is the damping. If the aim is to calculate the relative displacement within a door gap, then the stiffness of the seal is also of crucial importance to the accuracy of the results. To improve the input parameters attenuation and seal stiffness, both a standard modal correlation and a time domain correlation are carried out. This involves the exact and high spatial resolution measurement of the eigenmode of the complete model

as well as the relative displacements along the contact/gap lines between neighboring components. In the time domain correlation the relative displacement is analyzed according to the e-line method.

E-line method

The E-line method focuses on calculating and evaluating the relative displacement between two parts, which is the main cause for S&R, chafing and abrasive wear. The evaluation is always performed in the time domain and in a local coordinate system, in order to capture the displacement in the rattle direction and in the squeak plane (figure 1).

To enable an efficient evaluation of the relative displace-ment, node pairs are defined along a 3D curve, which is located between the two parts. Each node pair has its own local coordinate system, in order to capture the local gap geometry. The load is defined in time domain and can be taken from a PSD definition or recorded time domain data. The resulting displacements along all of the E-lines are calculated in the local coordinate system belonging to each node pair. The displacements are the input to an interface in Matlab, where the evaluation can be performed on a global, line and point level (figure 1, 3).

Since the result is a response in time domain (transient), a statistical approach is needed to include the time aspect in the evaluation. The amplitudes are ranked and a certain percentage of the highest values is chosen. Finally, the mean value of these amplitudes is calculated (figure 1). Hence, the whole time history of the relative displacements can be condensed into one single value, which can be compared to a tolerance value for squeak or rattle assessment.

1E-line along the tailgate gap (© Volvo)

4

Test

The measurements to determine the Eigen modes of the complete vehicle as well as the relative displacements along the joining points were carried out using a 'Robo-Vib' system. This measurement system enables an exact, non-reactive and high spatial resolution calculation of the 3D vibration movements of any structure. This is achieved through the combination of a six-axis robot and a Polytec 3D Scanning Vibrometer. Three measure-ment lasers are guided synchronously to the measure-ment points with the help of precision XY scanners. The transfer function is measured using the laser Dop-pler vibrometry procedure. The sensory function of the 'RoboVib' system is completely software-controlled, i.e. completely virtualized, and can be connected to the CAE world on data level. The measurement takes place directly on the simulation nodes – this simplifies and improves correlations significantly.

The RoboVib test system may be used for commercial measurement services such as, for example, experimen-tal modal tests with high point density.

Application test

This measurement was carried out in collaboration with engineers from Volvo Car Corporation and application engineers from Polytec. The tailgate in a complete, ready-to-drive Volvo V60 (figure 2) served as an example component for applying this method.

The vehicle is supported on tires with reduced air pres-sure in order to achieve a stiff body frequency. Vertical vibration is generated in the front left main chassis beam using a modal shaker and the force introduced is measured using a force transducer. First, the resonance vibration shapes of the entire vehicle are determined. To do this, a measuring grid that covers the entire vehicle shell is imported from the simulation (figure 3). The transfer function surface velocity/force is measured at every node. These results show the resonance vib- ration shapes at any given frequency, the high point resolution greatly improves visualization. Then the transfer functions are exported as a universal file and the Eigen modes and Eigen frequencies with the respective damping values determined in LMS using the Polymax procedure.

2Measurement set-up C (© Polytec)

3Measuring grid and deflection shape at 27 Hz (© Volvo)

5

The second step, which is included for frequency domain correlation, involves measuring the relative displacements of neighboring points on opposing sides along the tailgate gap. Since the points are to be measured sequentially, a pseudo random signal of 4 s - repeated for every measurement point (trigger) - is used as an excitation signal. The measured responses (movement in x, y and z) are saved as time data (figure 4). The underlying test procedure was carried out on an entire vehicle by the RoboVib Structural Test Station. Before, Volvo Car Corporation used to measure per 3D Scanning Vibrometer without a robot. Results are also taken from this test.

Simulation

The simulation model contains the BIW, lamps, bumper and tailgate. In order to bring the simulation model as close as possible to the test setup, the mass, geometry, material data and connections are thoroughly checked.As a means of further validation a visual modal correla-tion was carried out in order to compare the measured eigenmodes and resonant frequencies with those which were calculated.

The sealing is represented by spring elements, with stiffness in three directions following the sealing gap geometry, see figure 6. The spacing between the spring elements is 20 mm, which gives a length specific stiff-ness unit of N/mm/20mm for the sealing.

In order to calculate the relative displacement in exactly the same points as in the test, all the measured points

are exported from Polytec and imported into the simu-lation model. The E-line creation is performed by using these imported points. As shown in figure 5, the local z-direction is pointing in the gap direction.

A simulation in time domain is applied by using the modal transient analysis (SOL 112) in Nastran. The 4s long measured force signal from the test is used as an excitation for the simulation. Similar to the test, no boundary condition is applied (free-free). For each of the three test setups (without sealing, with sealing, with sealing and beam cross) a simulation has been performed.

5Seal represented with springelements – local gap geometry (© Volvo)

4Measuring grid with superim-posed snapshot from the relative displacement measured in the time domain (© Volvo)

6

Correlation

After calculating the relative displacement using the test and simulation, the results can be compared and analyzed in the Matlab program. Along the gap, the average of the 30 % largest values is used to enable a robust analysis. Before the test data can be compared with the simulation results, these must be filtered and, transformed from the overall test coordinate system into the corresponding local coordinate system for each node pair.

Figure 6 shows the magnitude of the relative displace-ment for the measurement (solid line) and the simu-lation without seal. A value of D = 1 % (dotted line) is assumed as an initial estimated value for the overall damping. Since the attenuation is the only unknown parameter, a good correlation can be achieved by re-ducing the damping value to D = 0.5 % (broken line).

The relative displacement for the local z-direction (rattle direction) is analyzed in the same way, figure 7. The broken line is the simulation response and the solid line the measured relative displacement. Both the shape and the values of both curves generate a good correlation in the local z-direction.

This good correlation is also achieved for the local y-('flush) and x-direction (along the gap). The damp-ing value of D = 0.5 % which was calculated in the first measurement is then used as the given value for the second measurement with seal. The three seal stiffness values are the unknown parameters in the second mea-surement. The values are varied independently of one another in the simulation. A change in one stiffness value changes the eigenmodes and resonant frequencies. This in turn leads to a change in the amplitude and phase in

6Relative displace-ment comparison between the test and simulation – the curve of the relative displace-ment starts in the top left corner and finishes at the top right corner of the tailgate(© Volvo)

7Relative displace-ment in local z-direction for test and simulation (© Volvo)

rela

tive

vibr

atio

n di

spla

cem

ent [

mm

]

E-Line pointE-Line point

rela

tive

vibr

atio

n di

spla

cem

ent [

mm

]

7

the modal transient analysis, making it more difficult to make a clear prediction. Solution possibilities are limited through a systematic variation of the parameters and analysis of the seal stiffness. A good correlation can be achieved with the following stiffness values: kx=5, ky=0, kz=3 N/mm/20 mm. The stiffness in the lateral direction (y) has a negligible influence on the relative displace-ment.

Conclusion

The validation on the overall damping and stiffness was conducted with the help of a correlation in the time domain. Both parameters are essential for squeak and rattle simulations of vehicles. Measurements were performed on the RoboVib Structural Test Station. The measurement results allow a very precise modeling of squeak and rattle behavior based on the exact simula-tion of the relative motion in all three local directions.

This article shows that numerical simulation and the described E-line method help to improve predictions about the acoustic behavior in compounds. The optimi-zation of a model requires both, a reliable methodology and the validation of the calculated results through realistic testing. The described e-line method has proven to be an efficient method.

The cooperation between Volvo Car Corpo-ration and Polytec RoboVib® Test Center has proven to be an efficient method.

Authors

Dipl.-Ing. Jens Weber, Technical SpecialistCAE Solidity, Volvo Car CorporationGothenburg (Sweden)

Dr.-Ing. Jochen Schell, Head of Applications, Optical Measurement Systems Division and Dipl.-Ing. (TU) Jörg Sauer, Vibrometry Product Manager and Strategic Product Marketing, Optical Measurement Systems Division, Polytec GmbH Waldbronn

OM

_AN

_VIB

_C_0

07_S

quea

k_Ra

ttle

_E_4

2460

20

18/1

0 - T

echn

ical

spe

cific

atio

ns a

re s

ubje

ct to

cha

nge

with

out n

otic

e.

Polytec GmbH (Germany)Polytec-Platz 1-7 76337 Waldbronn Tel. +49 7243 604-0 info@polytec.de

Polytec GmbH (Germany)Vertriebs- undBeratungsbüroSchwarzschildstraße 1 12489 Berlin Tel. +49 30 6392-5140

Polytec, Inc. (USA)North American Headquarters16400 Bake ParkwaySuites 150 & 200Irvine, CA 92618Tel. +1 949 943-3033info@polytec.com

Central Office1046 Baker RoadDexter, MI 48130Tel. +1 734 253-9428

East Coast Office1 Cabot Road Suites 101 & 102Hudson, MA 01749Tel. +1 508 417-1040

Polytec Ltd. (Great Britain)Lambda HouseBatford MillHarpenden, Herts AL5 5BZTel. +44 1582 711670info@polytec-ltd.co.uk

Polytec France S.A.S.Technosud IIBâtiment A99, Rue Pierre Semard92320 ChâtillonTel. +33 1 496569-00info@polytec.fr

Polytec JapanArena Tower, 13th floor3-1-9, ShinyokohamaKohoku-ku, Yokohama-shiKanagawa 222-0033Tel. +81 45 478-6980info@polytec.co.jp

Polytec South-East Asia Pte. Ltd.Blk 4010 Ang Mo Kio Ave 10#06-06 TechPlace ISingapore 569626Tel. +65 64510886info@polytec-sea.com

Polytec China Ltd.Room 402, Tower B Minmetals Plaza No. 5 Chaoyang North AveDongcheng District 100010 BeijingTel. +86 10 65682591info-cn@polytec.com

www.polytec.com