Simulating thrre directional vibration on a one-directional shaker

[CaptainCrunch]

Hello All,

I was hoping to get some thoughts on simulating three directional vibration on a one-directional shaker. I am running small engines (3-30 Hp, one to three cylinder) and engine components (mufflers, air cleaners, carberateurs, etc.) on 3500 lbf TIRA shaker. Generally I test 4-20 components at once, depending on size / wiehgt / complexity. I generally run time history data and accelerate it 2-3 times amplitude.

If the part is very simple I try to orient the components on the shaker in the acc. vector sum direction. But the fixtures get very complicated have muliple asemblies orienteed in some odd orientation.

Most times I run the three driections sequentially. (For example parallel to the crank, then perpendicular, then vertical). I try to find the dominant direction of vibration and replicte the field data in that direction. IF no failures occur on the first direction then I do the seoncd, and finally third direction. This matches the three direction of vibration measured at a point with a tri-axial acc.

So my main question is how accurate is it run these tests sequential versus orienting the component in the acceleration vector direction and accelerating it by the magnitude of the 3D acceleration?

Has anyone studied this issue?

Thanks in advance.

George Vanydke
Briggs & Stratton

 

[KalleH]

CaptainCrunch,

I guess that the simplest way for you to determine if your doing things right would be to intrument a specimen with strain gauges in the most critical points and obtain both strain and acceleration when you do your DAQ in the field. Then see if the same points will see the same damage during your test.

I have a strong hunch that you are not testing quite right however.

In cases where we only have one DOF to play with but I have an MDOF time history from a test track as load, we have been analysing the movements of our part as measured on the test track using ODS. Based on this we then:

(1) Orient the part so that we can get all the natural frequencies going with one loading direction.

(2) For the parts of the load that can be assumed to be stationary random, we use a shaped noise in the shaker.

(3) For the non stationary random parts we look at which natural frequencies we excite and design constant amplitude harmonic loads for the shaker.

Cheers!

 

[CESSNA1]

DEAR CAPTAINCRUNCH: Your method is the traditional way of doing three axis vibration testing. There is (was) only one outfit that I knew of that built 3 axis electrodynamic shakers. It is "Screening Systems Inc". I do not know if they are still around. Another way to do three axis vibration is with acoustics. One question: is three axis vibration really necessary? You may wnat to revisit the requirements. I am curious to know the reason.

Regards
Dave

 

[GregLocock]

You won't get the same fatigue damage, in theory, as a true 3 axis excitation. An obvious reason of this is when damage accumulates as strain^6, so if you had 2 equal components you'd only get 2*.7^6=24% of the real damage.



Cheers

Greg Locock