Natural frequency
ABSTRACT
In the automotive industry, experimental modal analysis is used extensively in new vehicle development to design
various components and assemblies for road dynamic loads. Experimentation is essential to bridge the gaps in
simulation. In this paper, we present three automotive components exhibiting increasing level of complexity in
their dynamic response. These automotive components contain some elements that are difficult to model because
of insufficient data available on material properties. The protocol adopted in each case is as follows: 1. Construct
a finite element model of the component to deduce modal parameters, namely, natural frequencies, and mode
shapes; 2. Experimentally determine the modal parameters of the system under test, including realistic natural
frequencies and damping coefficients; 3. Correlate experimental measurement with simulation results. These
simulation and experimental studies are carried out with free-free boundary conditions. Data acquisition for
experiment, modal parameter estimation, post processing and correlation are performed using an in-house
software package.
1 Introduction
Three automotive case studies are presented in this paper to illustrate the utility of Experimental Modal Analysis
in validating and subsequently updating Finite Element models. The cases chosen here are an automotive wheel,
an automotive suspension system and an exhaust pipe model of a car. These structures exhibit increasing
complexity in their material properties and dynamic response characteristics and therefore pose difficulties in
accurately modeling them using finite element simulation.
The wheel example is a component level study of a metallic rim to which refinements are made to its material
properties based on experimental results. The suspension system is a modal study of an assembly of
components having an operational degree of freedom between them. The challenge here is to appropriately
represent the assembly in the FE model and accurately model the degrees of freedom of the individual
components in relation to each other. The third case study of the exhaust pipe model describes a system for
which there is initially insufficient information about its material properties and how modal testing helps in refining
modeling assumptions to match the realistic experimental model more accurately.
The simulation procedure involves the construction of a geometric model of the test specimen, meshing the model
with appropriate mesh elements, selection of material properties and assigning boundary conditions. The solid
models are built in SolidWorks and the meshing and finite element solvers used are FEMAP and
NASTRAN respectively.
The experimental modal tests and analyses are carried out using an in-house vibration measurement and modal
analysis software called VMAPTM (Vibration Measurement and Analysis Package). The test procedure in VMAPTM
comprises of geometry creation and import, data acquisition from impulse hammer and shaker excitations, curve
fitting of measured frequency response functions for modal parameter estimation of natural frequencies, damping,
and amplitudes, post-processing and visualization of mode shapes, and correlation of FE results with test.
2 Car Wheel Modal Analysis
A modal analysis study is performed on a car
wheel freely suspended as shown in figure 1
and having the following dimensions:
Wheel Diameter = 0.365m
Thickness = 0.003m
Rim Width = 0.140m
2 c lo s e ly s p a c e d m o d e s
C M IF c u rv e
Figure. 5 MDOF curve fit using RFP method to identify first 2 closely spaced modes
In addition modal parameters were estimated from a nonlinear least squares curve fitting method to verify results
from RFP. The natural frequencies and Q factors were found to match closely for both these modes as is seen in
table 1.
RFP
n (Hz)
Q Factor
n (Hz)
Q Factor
Mode 1
267.1
898.1
268.1
1002.2
Mode 2
273.4
620.4
272.1
596.9
Mode 3
738.4
491.4
740.1
349.8
Table. 1 Natural frequency and damping factors estimated from 2 different MDOF curve fitting methods
2.3 Results and FE-Test Correlation
The first two closely spaced modes obtained from modal parameter estimation of the experimental data also
reveal two symmetric modes like in the case of the simulation model. They arise out of physical symmetry of the
wheel geometry. Thus while the mode shapes are similar for both modes occurring in two different planes of
symmetry; they are in reality two distinct modes with different resonant frequencies.
Figures 6 and 7 show a visual correlation of mode shapes obtained from the test and FE models. The mode
shapes are found to be similar for modes, 1 and 3. This brings us confidence in the correctness of assumptions
made in the simulation model.
Figure. 6 Comparison of mode shapes for mode 1 between EMA (left) and FEA (right) images in VMAPTM
Figure. 7 Comparison of mode shapes for mode 3 between EMA (left) and FEA (right) images in VMAPTM
3 Suspension System Modal Analysis
This experiment is an example of a modal analysis study conducted on an automotive suspension system
suspended in free-free condition as shown in figure 8.
Compressionmodeofspring
CMIF
FitCurve
Compression
Mode
n (Hz)
Q Factor
n (Hz)
Q Factor
30.4
147.3
30.3
161.9
Table. 2 Natural frequency and damping factors estimated from 2 MDOF curve fitting methods
4 Exhaust Pipe Modal Analysis
This case study involves the analysis of an automotive exhaust system as shown in figure 10. The pipe is a
combination of 3 units; the exhaust and engine manifolds connected to either side of a central drum. The
specifications of the test specimen are a total length of 1.3m and weight of 6kg. The entire system was
suspended by chords (as shown in figure 10) to simulate free-free boundary conditions.
The drum, like in a typical exhaust system, is
not completely hollow and is presumed to
comprise of a metallic mesh in the form of
filters or a catalytic converter and possibly
some filtrate spread across its inner surface.
Therefore, while the material properties of the
two manifolds were known, the primary
challenge in modeling the system was to
estimate accurately the density of the central
drum for the simulation model.
n = 55.74Hz
n = 60.37Hz
RFP
n (Hz)
Q Factor
n (Hz)
Q Factor
Mode 2
60.34
608.27
60.61
618.65
Mode 3
116.59
88.21
116.67
86.51
Table. 3 Natural frequency and damping factors estimated from SDOF and MDOF curve fitting methods
4.3 Results and FE-Test Correlation
A visual correlation of the mode shapes from test and FEA was performed in VMAPTM as shown in figure 13
below. It can be explained that the closely spaced modes, modes, 1 and 2 and modes, 3 and 4 exist because of
the near physical symmetry of the structure. For the same reason, identical mode shapes are also seen for the
two closely spaced mode pairs, the mode shape being repeated in the two planes of symmetry of the structure.