Optimizing Random Vibration Test Specifications for Shock Requirements

Dynamic Concepts, Inc. Huntsville, Alabama Vibrationdata Vibrationdata Using a Random Vibration Test Specification to Cover a Shock Requirement via a Pseudo Velocity Fatigue Damage Spectrum By Tom Irvine 3rd International Conference on Material and Component Performance under Variable Amplitude Loading, VAL2015 1

Vibrationdata Vibrationdata Introduction Shock Fatigue 1. Determine whether a given PSD can cover an SRS Specification 2. Derive an Optimized PSD which will cover an SRS

Vibrationdata Vibrationdata References H. Gaberson, Shock Severity Estimation, Sound & Vibration Magazine, Bay Village, Ohio, January 2012 H. Caruso and E. Szymkowiak, A Clarification of the Shock/Vibration Equivalence in Mil- Std-180D/E, Journal of Environmental Sciences, 1989 Dave Steinberg, Vibration Analysis for Electronic Equipment, Second Edition, Wiley- Interscience, New York, 1988 ASTM E 1049-85 (2005) Rainflow Counting Method, 1987 Halfpenny & Kim, Rainflow Cycle Counting and Acoustic Fatigue Analysis Techniques for Random Loading, RASD International Conference, Southampton, UK, July 2010 Halfpenny, A Frequency Domain Approach for Fatigue Life Estimation from Finite Element Analysis, nCode International Ltd., Sheffield UK

Vibrationdata Vibrationdata Electronics Solder Joints Aerospace and military components must be designed and tested to withstand shock and vibration environments Cracked solder Joints for Piece Part with J leads

Vibrationdata Vibrationdata Introduction Consider a launch vehicle component which will be exposed to random vibration and pyrotechnic shock during flight The random vibration occurs primarily during liftoff and the transonic and maximum dynamic pressure phases of ascent. The corresponding random vibration specification is in the form of a base excitation power spectral density (PSD) The pyrotechnic shock is due to staging and separation events, with the resulting shock requirement given as a shock response spectrum (SRS)

Vibrationdata Vibrationdata Shock & Vibration Testing Shaker Table Vibration Test Shock Testing using a Resonant Plate Usually straightforward to meet specification Typically excited by mechanical impact from pneumatic piston. Requires trial-an- error configuration to meet specification

Vibrationdata Vibrationdata Test Concerns Aerospace Pyrotechnic-type SRS tests are almost always more difficult to configure and control in the test lab and are thus more expensive than shaker table PSD tests Some lower and even mid-level SRS specifications may not have the true damage potential to justify shock testing The purpose of this presentation is to demonstrate a shock and vibration comparison method based on the fatigue damage spectrum (FDS) The comparison results can be used with other considerations to determine whether the random vibration test covers the shock requirement A related method is also demonstrated for deriving an optimized PSD to envelop an SRS These methods are found to be effective comparison and derivation tools within a framework of assumptions

Vibrationdata Vibrationdata Test Concerns Gaberson, et al, have characterized shock damage potential in terms of pseudo velocity A typical velocityseverity threshold is 100 in/sec (254 cm/sec) for military quality equipment some references apply a 6 dB margin which reduces this limit by one-half. This threshold is defined in part by the observation that the velocity which causes yielding in mild steel beams is about 130 in/sec Also note that some aerospace and military standards for electronic equipment define a shock severity threshold as 0.8 G/Hz times the natural frequency in Hz, which is equivalent to 50 in/sec References: MIL-STD-810E & SMC-TR-06-11

Vibrationdata Vibrationdata Test Concerns Shock tests may be omitted for some components if the pseudo velocity is < 50 in/sec The argument to skip shock testing can be strengthened if the random vibration test is rigorous enough to cover the shock requirement The study in this webinar uses numerical simulations to compare the effects of random vibration and shock via rainflow cycle counting and fatigue damage spectra The comparison can then be used with other factors to determine whether a random vibration test covers a shock requirement

Vibrationdata Vibrationdata Assumptions The component can be modelled as a linear single-degree-of-freedom (SDOF) system The peak shock and vibration pseudo velocity response levels fall below the threshold for the corresponding material, or below 100 in/sec for an electronic component The resulting shock and vibration response stress levels are below the material yield point Fatigue is the only potential failure mode The lower level, longer duration random vibration test may be effectively substituted for the high-amplitude, brief-duration shock test

Vibrationdata Vibrationdata Assumptions (cont) There are no failure modes due to peak relative displacement, such as misalignment, loss of sway space, mechanical interference, etc There are no shock-sensitive mechanical switches, relays or reed valves, which might experience chatter or change-of-state during shock There are no extra-sensitive piece parts such as crystal oscillators, klystrons, travelling wave tubes, magnetrons, etc The piece parts are Mil-spec quality and have been previously qualified to shock levels similar to those in MIL-STD-202, MIL-STD-883, etc The natural frequency, amplification factor Q and fatigue exponent b, can be estimated between respective limits

Vibrationdata Vibrationdata Rainflow Cycle Counting SDOF responses must be calculated for each fn and Q of interest, for both the PSD and the for SRS A representative time history can be synthesized for the SRS The Smallwood, ramp invariant, digital recursive filtering relationship is then used for the response calculation per Reference The rainflow cycles can be calculated from the time domain response In addition, response PSDs can be calculated for the base input PSD using the textbook SDOF power transmissibility function The rainflow cycles are then tabulated from the response PSDs via the Dirlik method

Vibrationdata Vibrationdata Fatigue Damage Spectrum A relative damage index can be calculated from the response rainflow cycles using = b i D A n i i 1 = The FDS expresses the damage D as a function of natural frequency with the Q and b values duly noted The amplitude convention for this paper is: (peak-valley)/2

Vibrationdata Vibrationdata Example Determine whether a given PSD envelops an SRS in terms of fatigue damage Natural frequency is an independent variable, 20 to 2000 Hz Vary amplification factor Q = 10 or 30 Vary fatigue exponent b = 4 or 9 The natural frequency, damping and fatigue exponent respective estimates are wide open because electronic boxes are typically black boxes for mechanical engineering purposes Wide estimates also allow for a rigorous test of the method.

Vibrationdata Vibrationdata PSD Specification Power Spectral Density, 24 GRMS Overall 1 Freq (Hz) Accel (G^2/Hz) Accel (G2/Hz) 20 0.04 0.1 150 0.30 2000 0.30 0.01 20 100 1000 2000 Frequency (Hz) Duration 180 sec/axis

Miscellaneous > Fatigue Toolbox > PSD Input > VRS & FDS for Base Input PSD Run this for all four (Q, b) permutations. Save each Pseudo Velocity FDS.

Vibrationdata Vibrationdata SRS Specification Natural Frequency (Hz) Accel (G) 10 10 2000 2000 10,000 2000 Three shocks/axis

Vibrationdata Vibrationdata SRS Specification Pseudo Velocity Shock Response Spectrum > Convert Accel SRS to Pseudo Velocity SRS

Vibrationdata Vibrationdata SRS Specification Pseudo Velocity Maximum PV = 61 in/sec

Synthesize a time history from scratch or use library file: srs2000G_accel Only need one time history because spec is always Q=10 even though two Q values are used for FDS

Vibrationdata Vibrationdata Synthesized Time History

Vibrationdata Vibrationdata SRS Specification Shock Response Spectrum Q=10 10000 Spec & 3 dB tol Negative Positive Natural Frequency (Hz) Accel (G) 1000 Peak Accel (G) 10 10 100 2000 2000 10,000 2000 10 510 100 1000 10000 Natural Frequency (Hz) Three shocks/axis

Run this for all four (Q, b) permutations. Save each Pseudo Velocity FDS.

Legend: PSD SRS Pseudo Velocity FDS Q=10 b=4 Pseudo Velocity FDS Q=10 b=9 1010 1018 Damage (in/sec)4 Damage (in/sec)9 109 1016 108 1014 107 1012 20 100 1000 2000 20 100 1000 2000 Natural Frequency (Hz) Natural Frequency (Hz) Pseudo Velocity FDS Q=30 b=4 Pseudo Velocity FDS Q=30 b=9 1011 1020 Damage (in/sec)4 Damage (in/sec)9 1010 1018 109 1016 108 1014 20 100 1000 2000 20 100 1000 2000 Natural Frequency (Hz) Natural Frequency (Hz) PSD Covers SRS for b = 4 (plots in left column)

Vibrationdata Vibrationdata SRS Specification Now consider the case where a PSD is to be derived to cover an SRS requirement. The component will be assumed to have Q=30 and b=6.4 (single pair for brevity) The natural frequency is left as an independent variable. Candidate PSD functions can be derived via trial-and-error Each PSD is scaled so that its pseudo velocity FDS just envelops that of the time history synthesized for the SRS specification The optimal PSD is that which satisfies the enveloping with the least possible acceleration, velocity and displacement RMS levels

Time History > PSD Envelope via FDS

Power Spectra Density 47.2 GRMS Overall, 180 sec Pseudo Velocity FDS Q=30 b=6.4 1016 10 SRS PSD 1015 Damage (ips)6.4 1 Accel (G2/Hz) 1014 0.1 1013 1012 0.01 20 100 1000 2000 20 100 1000 2000 Frequency (Hz) Natural Frequency (Hz) Freq (Hz) Accel (G^2/Hz) The equivalent PSD is conservative in terms of fatigue damage. 20 0.026 137 0.65 2000 1.476

The equivalent PSD does not completely envelop the SRS. Increase the level or duration if peak enveloping is required.

Vibrationdata Vibrationdata Peak Enveloping A conservative PSD can be generated to envelop an SRS in terms of peak response But PSD is limited to about 2000 Hz for practical shaker test This limitation is okay as long as component is an SDOF system with fn < 2000 Hz

Shock Response Spectrum > Envelope SRS via PSD, peak response

Vibrationdata Vibrationdata Peak Envelope PSD But too high for a shaker table test!

Vibrationdata Vibrationdata Comparison The peak VRS is based on the Rayleigh distribution.

Vibrationdata Vibrationdata Conclusions Rainflow FDS curves can be calculated for both PSD and SRS functions The curves can then be superimposed on the same graph to compare the damage potential for each environment The relative differences between the FDS curves for the PSD and SRS for the first example were rather insensitive to Q but very sensitive to b The FDS comparison technique can also be used as a basis for enveloping a shock event with a PSD optimized in terms of the least possible overall levels, as shown in the second example These methods can be used more efficiently if the natural frequency, damping and fatigue exponents respective estimates can be narrowed Matlab scripts for performing these calculations are available at: https://vibrationdata.wordpress.com