By: George Henderson President GHI Systems, Inc. This material is protected by the copyright laws of USA. Any reproduction of this material is prohibited.

By:

George HendersonPresident GHI Systems, Inc.

This material is protected by the copyright laws of USA.Any reproduction of this material is prohibited by law.

Original Paper Presented at the 2004 IEEE-AST

Conference in Chicago, Oct. 2004.Revised Version Released Jun, 2005

Proving HALT/HASS Machine Effectiveness Using Fatigue Spectrums

And A Method Of Machine Fatigue Comparison


2

What some HALT/HASS Machine manufacturers won’t tell you

Discussion of problem – “what is impulsive vibration”?

What does Grms measure?

The fatigue metric machine comparison

Conclusions - References

Introduction


3

Hammer excited 6DOF machines can have significant variations in Grms intensity at different table locations.[1,2] This is obviously due to differences in the PSDs across the table surface.

Intensity Variability Problem

24 Sites, bars heights equal Grms amplitude.

Z Axis – Red,

X Axis – Blue,

Y Axis – Green

Data Courtesy HP Greeley


4

Examples of Variation Magnitude◊ Uneven fatigue results from uneven excitation

◊ BP = Before installation of heavy plate on table. AP = After installation of plate.


5

Example of Structure Transmission Loss

UUT was a 20 lb medical patient monitor with rubber feet resting on shaker table and fixtured to the table with hold-down bars.

It Will Be A Problem Also.


6

6DOF and ED Shaker PSD’s

RED = 6DOF @ 10Grms

BLUE = ED NAVMAT @ 6gGrms

Equality @ 500Hz


7

The Grms Issue

g2/H

z Moderate Grms

◊ Both PSDs have the same area, hence the same Grms. ◊ The stress for case A is lower then case B for the same fr. PSD B will be more damaging @ fr than will be PSD A. But it’s the g2/Hz power magnitude @ fr that results in fatigue.

◊ Grms is unrelated to stress, as illustrated above.

Case B

Consider Two Arbitrary PSDs having the same area


8

Grms and AcresAcre – Regardless of it’s shape or qualitative value, the standard measure of an area of land.

Grms - Regardless of it’s shape or qualitative value, the standard measure of the area of a PSD.

The wise farmer places secondary significance in the magnitude of acres since it

lacks qualitative value.

Likewise, the 6DOF machine user should understand that Grms gives no qualitative value of the fatiguing potential of a vibration.


9

What Is Impulsive Vibration?• For impulsive and displacive excitations, fatigue results from repeated stress loading cycles at the component’s self-resonant frequency fr.

• Remember “Is it real or is it Memorex?”

•Uneven stress loading produces EOL implications.


10

A Note on Vibration Induced Fatigue

Contrary to popular belief,

Fatigue is proportional to:

First bending mode velocity –

not acceleration.

Reason: Strain results from displacement, and Strain cycles equate to fatigue.


11

Why Is A Velocity Spectrum Needed?

• Fatigue relates to the velocity of the 1st self resonant vibration bending mode, not acceleration.[4,5] To analyze it, it is best to express the velocity of vibration in the form of a spectrum of velocity frequencies.

– A metric for its determination, the Damage Potential Spectrum, (DP(f), was introduced in 1995.[3]

– The DP(f) Is a velocity spectrum, It Is derived from the acceleration PSD using fatigue “operators.”

– Operators are the duration of applied stress T, vibration element damping ζ, and S/N fatigue constant, β of the damped element.


12

The DP(f) Equation’s Relation to PSD

• The DP(f) is derived from the PSD.– As such it is easier to understand.

• It differs from the PSD in that it is a velocity spectrum indicating fatigue potential, not spectral vibration power.

• It uses ‘operators’ to adjust for fatigue. Operators are:– Stress loading time (T), damping (ζ), fatigue slope (S/N)

• It is simply scaled in relative fatigue potential.


13

What is DP(Avg)?

• It is the average spectral intensity over a defined velocity spectrum bandwidth.

• It gauges the overall intensity of fatigue accumulation potential over the bandwidth.

• In general, a spectrum with higher DP(Avg) produces fatigue more rapidly than a spectrum with lower DP(Avg).


14

DP(f) provides a method of characterizing HALT-HASS shakers in terms of Fatiguing Potential.

This presentation describes the DP(f) method of characterization of HALT/HASS machines.


15

An Experiment

•A Controlled Experiment Was Performed at a HALT Facility Having Two 6DOF Machines.

– Machine #1 - 26” x 30” W/6 Hammers.– Machine #2 - 20” x 24” W/4 Hammers.

•Data Was Taken from Two Z Axis Accels @ Identical Table Quadrants on Each Machine.•Both Controllers Were Set At 10 Grms


16

DP(f) Fatigue Factors Used• Fatigue Constants Were:

– Duration = 60 minutes, Mean Damping Ratio ζ = 5%, Mean S/N β = 4.

– ζ and β values are for Typical Vibration Testing as Recommended for Electronic Assemblies by MIL STD 810F-514.3 Vibration.

• DP(f) Spectrums from 20 Hz – 2 KHz Were Estimated.– These Were Over-laid in Order To See Differences


17

Typical Overlaid DP(f) Spectrums


18

Experiment Summary

• Expected: …………Machine #1 More Damaging.• Actual: ……………..Machine #2 More Damaging.

• DP(Avg) #1 = 17.886 Fatigue Units• DP(Avg) #2 = 36.396 Fatigue Units

• DP(f) Fatigue Ratios @ Specific fr Exceed 500:1.

• No Relationship of Control Grms to Fatigue.• Results suggest method of establishing machine

effectiveness for comparative purposes.


19

Comparing Fatiguing Power Of Shakers

• A variant of the previously described Experiment was developed.

• Instead of only two, a systematic method using at least four measurement locations was established. The more locations, the better.

• DP(Avg) statistics are used.


20

Characterization Method

• Divide table into equal area zones.– Perhaps zones enveloping fixturing points.

• Instrument zones, either Z axis or triaxial sensors.– The Qualmark “Tripod” may be used.

• Run chamber @ ‘Standard’ control point ~ 10 Grms.• Obtain DP(Avg) fatigue data for each zone.


21

Method Details

• Let shaker stabilize before measurements.• When taking data, don’t change setpoint Grms.• Average spectrums at least 5 - 10 times.• Zoom spectrums - 100 Hz to 3 KHz suggested -

should be set by UUT characteristics, but be consistent!

• Obtain DP(Avg) from each table position. – (Similar to Grms of PSD).

• Statistically analyze DP(Avg) values.


22

Four Zone DP(f) Plots


23

Zoomed DP(Avg) Data ValueZoomed portion of spectrum – DP(Avg) is from this portion

of spectrum .


24

Comparative Analysis• Estimate MEAN and STANDARD DEVIATION of resulting

DP(Avg) values:

– Mean = 22.37 - STANDARD DEVIATION = 5.64• MEAN is average level of fatigue potential Effectivity.• STANDARD DEVIATION is range of non-uniformity.• More zones produce more accurate Means and SD.

Zone # 1 2 3 4

DP(Avg) 21.12 21.32 31.34 15.72


25

Adjustment Methods

• A shaker may be adjusted by “compensation” to raise or lower average fatigue values.

• 1) Fatigue is proportional to duration of applied stress.– Adjust screen times to achieve desired fatigue totals.

• 2) Fatigue is indirectly proportional to Grms magnitude.– Adjust Grms to effect desired fatigue level @ stress time

T.• These general rules relate to average situations, actual DP(f)

values should be used for specific test object analysis.


26

“Single Spectrum” Comparisons• Less accurate – greatly influenced by table hot\

cold excitation spots.• Earlier described “Experiment” was of this type.• Not satisfactory for machine to machine

comparison.– Statistical analysis of more data points

provides vastly improved accuracy.• Produces invalid competitive comparisons of

one machine or company versus a second. • Don’t waste your time on this or the use of Grms

to define a shaker. A table survey is required.


27

A Case History Of Machine Differences

• Two different brand name machines• Screening the same part• Parts had identical flaws

– Through hole mounted computer crystal can

• Controls set at the same 10 Grms• Vastly different fatigue results• PSD levels greatly different at crystal resonance


28

The Lincoln Computer Case Study

HALT Tests at Ford Lansdown and Detroit.

Identical flaw in Lincoln ‘Ride’ Computers, Same fixture, Same Grms.

Flaw was THM Xtal with High Standoff.

Excitation at fr of Xtal Caused Fatigue Failure.

Difference was g2/Hz Intensities at fr of Xtal.


29

Simplified PSDs Of Shakers

Frequency - Hz

10 100 1000 10000

Sp

ectr

al I

nte

nsi

ty g

2/H

z

Self-resonance fr of Xtal

~ 420 Hz

–Machine 1

–Machine 2


30

Fatigue vs Time Diagram

Time - Minutes 1 10 100 1000

Acc

um

ula

ted

Fat

igu

e > Xtal Lead Fatigue Destruct Level –

8 min 55 min

Mean fatigue Time-To-Failure diagram Avg of 10 samples/machine

–

Mac

hine

1

–

Machine 2


31

Recommendations

Recommendations for 6DOF users:

Don’t assume:– Setpoint Grms correlates to fatigue rate, there is

no direct correlation.– Fatigue is uniform at all locations on a table.

It’s not!

Apply “Hanks Rules” for more successful HALT-HASS results.


32

Conclusions• Grms does not indicate fatigue fate –

– Fatigue is not related to vibration acceleration.

• Variations in g2/Hz and fatigue potential are relevant.

• At identical Grms settings, similar 6DOF chambers don’t have equal fatigue potential.

• DP(Avg) gauges table zone spectrums and overall machine fatigue producing potential.


33

Key References1. Henderson, G., “Dynamic Characteristics of Repetitive Shock Machines,” Proc.,

39th Annual Technical Meeting of IES, Vol. 29, No. 10, pp. 232-249, 1993.2. Henderson, G., “RS Machine Dynamics,” Accelerated Testing Forum,” Boeing

Commercial Airplane Group Conference, May 8-9, 1996 3. Henderson, G., and Piersol, A. B., “Fatigue Damage Related Descriptor for

Random Vibration Test Environments,” Shock and Vibration, Dynamic Testing Reference Issue, pp. 20-24, October, 1995.

4. Crandall, S., “Relationship between Stress and Velocity in Resonant Vibration,” Journal of Acoustical Society of America. Vol. 34, No. 12,pp. 1960-1961, 1962.

5. Gaberson, H. A., and Chalmers, R. H., “Modal Velocity as a Criterion of Shock Severity,” Shock and Vibration Bulletin, No. 40, Pt 2, pp. 31-49., 1969.

A complete list of References is available from the speaker.


34

The GHI Systems ProCAT

The 8 channel Spectrum Analyzer used for DP(f) analysis in this presentation, was

The GHI Systems ProCAT,

available from QualMark or directly from GHI Systems, Inc.


35

Thank [email protected]

© 2004-2005

GHI Systems, Inc. San Pedro, CA 90732

All Rights Reserved

By: George Henderson President GHI Systems, Inc. This material is protected by the copyright laws of USA. Any reproduction of this material is prohibited.

Documents