Random Vibration Fatigue Analysis

[Syed7777777]

Hi Everyone,

I'm trying to do a "Random Vibration Fatigue Analysis" on a bracket assembly that is mounted on the headboard of a pickup truck.

BACKGROUND:

The bracket assembly is mounted on the headboard of the truck using 2xbolts (1 per each mount). The truck regularly travels on a rough rocky terrain that causes a lot of vibrations. The bracket assembly has been failing before its expected life.

I'm trying to simulate this in FEA to identify the weak spots, compare the life suggested by FEA to what we're actually getting, modify the design accordingly, and then re-do the analysis.

HERE'S WHAT I'VE DONE SO FAR:

We bought a couple of tri-axial accelerometers and fixed them at the two(and only) mounts of the bracket assembly. Drove the truck through the rough terrain and recorded the vibration time-history data of both mounts in x, y and z directions.

After that, I converted each of the six 85-minute long time histories into FFTs and then converted FFTs to PSDs using DewesoftX.

Then using Abaqus, in the first step I ran a frequency analysis and extracted natural frequencies and mode shapes of the bracket assy. In steps 2-4 I ran 3x steady-state dynamic(modal) analyses in x, y and z directions with a "Unit Load Curve" of 9.81 amplitude to represent gravity.

I also selected the outputs for "Generalised Displacements and Phase Angles" which are used by Fe-safe.

I then imported the odb file in Fe-safe using the option of "Open Finite Element Model for PSD Analysis", imported the PSD file, and selected Polar(degrees).

After defining the materials to the groups, you get an option to define "length per repeat in seconds".

QUESTIONS:

1. Should I put 85x60= 5100 seconds in length per repeat?

2. If yes, then how do I relate or extrapolate that to get the actual(total) life of the bracket assembly if the truck travels through that terrain(represented by the 85-minute time history) 5 times a day, every day of the year?

3. Am I doing everything correctly?

4. Is there an easier/ better alternative to what I'm trying to achieve?

I would really appreciate elaborate and easy-to-understand answers without jargon.

Thanks. :)

 

[jrmichler]

Is there an easier/ better alternative to what I'm trying to achieve?

I would really appreciate elaborate and easy-to-understand answers without jargon.

You don't say, but I assume that the bracket is supporting something heavy. Otherwise, it would not break.

You don't say whether this is a one off home built bracket for a personal project, a limited production aftermarket bracket, or a high production factory bracket. It would help if we knew which it is because that affects our recommendations.

I'm trying to simulate this in FEA to identify the weak spots

No need for FEA to do that. The weak spots are where it breaks.

We bought a couple of tri-axial accelerometers and fixed them at the two(and only) mounts of the bracket assembly.

This tells you the vibration excitation to the base of the bracket. It does not tell you the peak stress in the bracket.

The truck regularly travels on a rough rocky terrain that causes a lot of vibrations.

Is the bracket in a place where a person can observe it while driving through the rough terrain? If so, does the bracket visibly flex? If so, no analysis is necessary. Just observe where it is flexing, and stiffen those areas until the bracket no longer visibly flexes.

A slightly more sophisticated approach is to observe the worst case direction and amount of flexing, then apply a static load to the FEA model to get that amount of deflection. Strengthen the high stress areas until the total deflection decreases by at least half. Build and test the redesigned bracket. A general rule of thumb is that the bracket needs to be stiff enough that there will be no visible flexing or bouncing under the worst driving conditions.

More sophisticated: Look at the frequency distribution of your vibration data. There should be peaks at the 1 to 1.5 Hz whole body natural frequency, and the 10 Hz or so tire / suspension natural frequency. Most of the energy will probably be in that frequency range. The natural frequency of the bracket with its load attached should be higher than that frequency range.

 

[Juanda]

Here is a related article that might be useful. It covers random vibration analysis in combination with fatigue analysis (Palmgren-Miner) using the results previously obtained from the software.
It's using a different software than the one you mentioned but the fundamentals are the same.

https://iberisa.wordpress.com/2021/01/17/random-vibration-con-femap/

The article is in Spanish but I'm sure Google can easily translate it for you if necessary.

 

[FEAnalyst]

I've already replied to the OP's questions about fe-sage usage on the eng-tips forum so let's see if there are any follow-up questions here or there.

 

[Jone Smith]

hi
We need to see what is the range of your accelerometers?
This is very important. What is the lowest and highest frequency range that they can see?
If you use accelerometer sensors, you can only record high frequency vibrations. If you use a velocity sensor, you can see a wider range of vibrations. According to the type of movement you described, it seems that observation of vibration is better than the speed of vibrations. We did the same test in [spam link removed]. But our goal was to calculate the natural frequencies of the equipment parts.