You are on page 1of 8

Modal Damping Estimates from Static Load-Deflection Curves

Gary C. Foss
Structural Dynamics Laboratory
Boeing Commercial Airplane Group
Box 3707, MC 1W-06
Seattle, WA 98124


Abstract
Wire rope isolators are used extensively in industry for vibration and shock isolation. They are favored for their
relatively low cost and high damping capacity. Vendors offer nominal stiffness properties and average damping
values, but do not publish damping variations with displacement amplitude, preload, or cycle rate. An
investigation was performed to verify and extend the vendors data on several large isolators. Because of the
high spring rates, directly measuring the dynamic behavior was difficult, requiring large masses and high
excitation forces. An alternative method was used to measure damping from quasi-static load deflection curves.
This was compared with damping estimates from frequency response functions. Damping results from the load-
deflection technique showed significant variation with deflection amplitude.

Introduction
Wire rope helical spring isolators are used to isolate vibratory equipment from floors, buildings, or support
structures, or alternately to isolate sensitive equipment from vibratory floors or structures. They are typically
constructed by winding a length of wire rope helically between two pairs of clamping bars. One set of clamps
mounts to the equipment, the other set mounts to the support frame or floor. The isolators are designed to
support the static weight of the equipment and allow some additional amount of displacement which flexes and
slides the strands of the cable against each other, dissipating friction as heat. They are popular for reasons of
high damping effectiveness, low cost, low maintenance, thermal insensitivity, and durability. Figures 1 and 2 show
wire rope isolators of different sizes.


Figure 1 Figure 2

Stiffness
The stiffness of helical spring isolators is difficult to characterize for predicting dynamic performance. The spring
rate is not constant, but highly dependent on the changing geometry of the helix, the direction of loading, and the
amount of preload. Vendors may report stiffness as an average load displacement curve [1], or a pair of average
values, one for shock and another for steady state vibration [2].

Designers and analysts requiring more complete load deflection (L-D) information may have to make their own
measurements. Figure 3 is an L-D curve, measured in the lab for the isolator shown in Figure 1. Since these
springs are normally used with a compressive preload, compression is shown in the positive direction. The curve
shape is typical; the spring rate greatly stiffens when stretched in tension as the coils are pulled tight. In
compression, the rate softens slightly as the coils bow out. Then the rate stiffens again as the isolator nears a
bottom out condition. The hysteretic loop shows that the spring rates during loading and unloading are similar
(except near the extremes), but during reversal, the rate momentarily rises significantly.




Figure 3: load deflection curve for the helical isolator shown in figure 1

Another problem for designers is that stiffness under the same loading can vary from spring to spring due to
manufacturing tolerances. Two groups of four springs with the same model number were tested in this study and
the variation in spring rates was as much as 20%.

Damping
Since the purpose of helical spring isolators is to absorb energy, damping is a property of great interest. It is,
however, a challenge to measure, especially on larger springs. When asked about damping, vendors are
reluctant to be too specific. At least 10% might be a typical response.

The flexing of wire rope involves both coulombic and viscous damping. At very low vibration levels, the wire
strands stick together and little sliding occurs. The damping is low and the behavior is viscous. With higher
displacements, coulomb damping predominates as the wires break free and start to slide against each other,
absorbing large amounts of energy. At large displacements the bending and stretching of the wire strands
overshadows the sliding friction and viscous behavior again starts to show.

Damping is a dynamic property, and its measurement is most frequently associated with the fitting of a second
order equation of motion to measured frequency response functions of a dynamic system. This might be as
simple as finding halfpower bandwith points, or as comprehensive as solving a matrix eigenvalue problem. Both
approaches assume linear behavior in terms of spring rate and modes of energy dissipation. The helical spring
violates both assumptions.

To perform a dynamic test, the helical spring must be loaded with a mass and excited by some means such as an
impact hammer or electrodynamic shaker. Owing to the high damping exhibited by helical springs, testing of
large ones may require very high forces to excite the system through a full range of amplitudes.

An alternative to the dynamic measurement of damping is available if the spring can be slowly loaded while
measuring displacement.

Hysteresis
Hysteresis has been described as the lagging of an effect behind its cause [3]. In mechanical terms, the response
of a system to a load depends not only on the present magnitude of the load, but also on the previous history of
the load. Expressed mathematically, the response to the load is a double-valued function; one value applies when
the load is increasing, the other applies when the load is decreasing. . For example, Figure 4 (from the catalog of
hysteresis models, #23, Bouc-Wen [4]) shows the load deflection diagram of a nonlinear spring with hysteresis.
X is applied load and Z represents displacement. One complete cycle of loading and unloading is shown and the
area inside the loop is proportional to the energy dissipated as heat. [5]




Figure 4: Typical hysteresis loop Figure 5: Setup for isolator L-D testing

Damping From Hysteresis
An isolator was set up for quasi-static load deflection testing with a hydraulic cylinder, load cell, and displacement
transducer as shown in Fig 5. Static testing was performed to obtain L-D curves over the maximum design range
(for example, Figure 3). To characterize the hysteresis properties, the isolator was loaded in tension and
compression at a slow, fixed frequency, from zero up to a maximum displacement. In this case the cycle rate was
0.5 cycles per second. (The rate should be slow enough that inertial forces on the load cell are negligible.) Figure
6 shows a typical time history of displacement, for which the X scale represents sample number. In this case the
amplitude was increased by hand, but could have been done programmatically.


Figure 6: Displacement time history.

When force is plotted against displacement, the characteristic hysteresis loops shown in Figure 7 are obtained.
One way to describe this figure qualitatively is that it possesses two stiffness backbones. Positive compressive
loading proceeds along the upper backbone. At cycle reversal, the curve drops down and returns along the lower
backbone. J ust beyond the point of cycle reversal, the loading diminishes but the displacement lags, resulting in
momentarily high stiffness. This temporary stiffness might be as high as an order of magnitude greater than the
nominal stiffness.

Another feature worth noting is that low amplitude cycles occurring near the origin may be too small to reach the
backbone. This would be the amplitude range where little or no sliding friction occurs, the hysteresis loop
begins to close, damping is lower, and the average stiffness slope is much steeper.


Figure 7: Load-deflection paths for cycles of increasing load; a typical cycle is outlined in red.

If we take this series of load deflection cycles and measure the area inside each loop, we can obtain a damping
estimate for each loop. An example is shown in Figure 8 for the loop identified by the red trace in Figure 7.


Figure 8: Calculating the area of each loop

The area in yellow represents the energy lost during one complete loading and unloading cycle. The area of the
green ellipse of width and height represents the maximum potential energy for this load and deflection
path. If is the area of the hysteresis loop, the loss factor [6], is the ratio,
0
2D
0
2F
loop
A
0 0
D F
A
loop

= (1)
where the denominator represents the area of the green ellipse. A MATLAB script was written to go through the
data file and identify each loop, characterized by starting and ending points. The area inside these points was
computed with the POLYAREA command. From the set of these and the set of ellipse areas, loss factors for
each loop were tabulated. The percent critical damping (zeta) is half the loss factor and is plotted in Figure 9 for a
helical spring isolator of the type shown in Figure 2. An interesting feature of this curve is that the peak occurs
near the region in Figure 6 where the F-D cycles join the stiffness backbone.

helical isolator damping
0
5
10
15
20
25
0.001 0.01 0.1 1
disp, in pk/pk
d
a
m
p
i
n
g
,

%

Figure 9: Variation of damping with displacement, from hysteresis

Effect of Cycle rate
To see if this process was sensitive to cycle rate, it was performed at rates of 0.1, 0.5, 2.5, and 5 Hertz. The
hydraulic actuator and servovalve limited the maximum cycle rate. There appeared to be little or no sensitivity to
cycle rate, at least at these rates. Results are shown in Figures 10-13


CB17 Helical Spr ing 0.1 HZ Cycle
Rat e
0
10
20
30
40
0.001 0.01 0.1 1 10
Displacement , in. pk/pk
D
a
m
p
i
n
g
,

p
e
r
c
e
n
t
CB17 Helical Spr ing 0.5 HZ Cycle
Rat e
0
10
20
30
40
0.001 0.01 0.1 1 10
Displacement , in. pk/pk
D
a
m
p
i
n
g
,

p
e
r
c
e
n
t

Figure 10. Figure 11.
CB17 Helical Spr ing 2.5HZ Cycle
Rat e
0
10
20
30
40
0.001 0.01 0.1 1 10
Displacement , in. pk/pk
D
a
m
p
i
n
g
,

p
e
r
c
e
n
t
CB17 Helical Spr ing 5.0 HZ Cycle
Rat e
0
10
20
30
40
0.001 0.01 0.1 1 10
Displacement , in. pk/pk
D
a
m
p
i
n
g
,

p
e
r
c
e
n
t

Figure 12. Figure 13.







Effect of Preload
Since most applications have the helical spring isolator partially loaded and compressed with the weight of
equipment, it was necessary to see if preload affected the damping to any degree. A test was run with 2000 lb of
preload, processed and plotted in Figure 14. Effect of preload was judged to be minimal.

Helical Spr ing Isolat or CB1800-17 w/2000 lb
pr eload
0
10
20
30
0.001 0.01 0.1 1
Displacement , in pk/pk
D
a
m
p
i
n
g
,

%

Figure 14. Effect of preload

Comparison with Dynamic Data
The setup for dynamic testing is shown in Figure 15. Eight thousand pounds of mass was placed on four helical
isolators mounted to a structural floor. The stiffness of each isolator was about 12,000 lb/inch. The mass was
driven by four shakers with a total force capacity of 3600 lb. Frequency response functions (FRFs) were
measured at constant force and displacement sine sweeps, performed over the vertical spring/mass resonant
frequency. With the high damping, the maximum force was only able to achieve displacement of about 0.08 inch,
about 5% of the maximum displacement range for this isolator.


Figure 15, Dynamic test setup Figure 16, FRFs for different amplitudes

Helical Spr ing Damping f r om Sine Sweeps
0
5
10
15
20
25
0.001 0.01 0.1 1
max displacement , in. pk/pk
D
a
m
p
i
n
g
,

%

Figure 17, Damping values from Fig. 16

Figure 16 shows superimposed FRFs for displacements ranging from .0023 to .0793 inches p/p. Damping
estimates were obtained from rational fraction polynomial curve fits of these curves, plotted in Figure 17. The
data would appear to approximately confirm the left half of the curve in figure 9, measured with the hysteresis
technique. The curve fits at higher damping were more difficult to perform, owing to the skewed FRF shape and
implied non-linearity.

Other Isolator Types
As a check on the test setup for the hysteresis measurements, several other isolator types were installed and
characterized. A steel spring element would be expected to have only the very low hysteresis due to material
damping of the steel itself. A elastomeric spring should show more damping, but be more linear in its spring rate
than the helical isolator.

Damping of a Steel Coil Spring
A steel coil spring was tension loaded in the static test setup to about half its fully elongated range. It was
subjected to cyclic loading increasing from zero to about 500 lb. The data file was processed with the MATLAB
script used for the helical isolators. As expected, the difference between the loading and unloading paths was
very small, as evidenced by Figure 18. Figure 19 shows the cycle outlined in red in Figure 18. Figure 20 shows
the damping relationship after all cycles are processed.

-0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
100
150
200
250
300
350
400
450
load deflection path
disp, in.
F
o
r
c
e
,

lb
.
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
150
200
250
300
350
load deflection path for one cycle
disp, in.
F
o
r
c
e
,

lb
.

Figure 18. L-D cycles for a steel coil spring Figure 19. Typical cycle extracted from Figure 18.


. Fig. 20. Damping from all cycles

The higher damping at the lower displacements in Fig 20 probably results from small sources of joint friction in the
setup becoming more important as material damping in the steel diminishes.

Damping in a Rubber Vibration Isolator
A sample rubber equipment mount was obtained from a vendor and loaded into the static test setup. It was also
preloaded realistically, and subjected to cyclic loading. The data file was again processed with the MATLAB
script used for the helical isolators. Figure 25 shows the mounted strut and Figure 26 shows the cyclical
hysteresis loops. Figure 27 shows a typical loading cycle and Figure 28 shows the damping dependency with
displacement.


-0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
0
2000
4000
6000
8000
10000
load deflection path
disp, in.
F
o
r
c
e
,

l
b
.

Figure 25, Rubber vibration isolator Figure 26, Hysteresis loops

-0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06
2000
3000
4000
5000
6000
7000
8000
9000
load deflection path for one cycle
disp, in.
F
o
r
c
e
,

lb
.

Figure 27, Typical cycle from Figure 26 Figure 28, Damping from all cycles


Conclusion
A quasi-static alternative method to measure loss factor and damping was explored for a large helical spring
isolator, for which the conventional dynamic approach was not practical for large amplitudes. Large variations in
damping with displacement were observed. Where the methods could be compared at low amplitudes, they
seemed to agree. The technique was also applied to several spring elements where the expected load-
displacement relationships were confirmed.



References
[1] Catalog and design manual, Enedine Inc.
http://www.enidine.com/pdffiles/WireRopeCatalog.pdf
[2] Helical wire rope catalog, Aeroflex Corp.:
http://www.aeroflex.com/products/isolator/datasheets/cable-isolators/helical.pdf
[3] Hysteresis definition,
http://www.answers.com/topic/hysteresis
[4] Catalog of Hysteresis models
http://www.civil.canterbury.ac.nz/ruaumoko/hystresis.html
[5] Vibration Damping, Nashif, J ones, Henderson, 1985 Wiley, pg. 75
http://www.wiley.com/WileyCDA/WileyTitle/productCd-0471867721.html
[6] Viscoelastic Damping 101-Insight, Paul Macioce, Sound & Vibration, April 2003
http://www.roushind.com/news_downloads/white_papers/Insight.pdf

You might also like