ModeLAB

ModeLab is a MATLAB toolbox for performing modal analysis using experimental FRFs from roving hammer tests. This essentially recreates the functionality of modalfrf and modalfit in the Signal Processing toolbox but is FOSS.

A function called read_signalcalc has been included to read in ASCII files from DataPhysics SignalCalc, but this toolbox could be used with other software too with a custom defined file reader.

Installation

Download the latest release and add the directories to your MATLAB path.

Usage

1. Copy files

Copy the ASCII directories from SignalCalc to an appropriate location, which will be referred to "C:\path\to\root\of\signal_calc\data". Also copy the setup.csv file from the ModeLAB folder to your new directory.

Your directory should now have the following structure:

├── ASCII00001     
|  └ Hxysv00001.txt        
├── ASCII00002     
|  └ Hxysv00002.txt       
|
.
.
.
└ setup.csv

Note that the Run numbers do not matter (ie, it's fine if you skip some runs so the runs are numbered 1,2,6,10...). However, the order of the runs must be consistent with their ordering in the setup.csv from step 2.

2. Update setup.csv

Update the setup.csv file to reflect the setup of the roving hammer test. This has the following format:

```
xHammer,0,0,0,0,0,0,0,0,0
yHammer,0,0,0,0,0,0,0,0,0
zHammer,0.1,0.25,0.375,0.5,0.65,0.8,0.925,1.05,1.2
FxHammer,1,1,1,1,1,1,1,1,1
FyHammer,0,0,0,0,0,0,0,0,0
FzHammer,0,0,0,0,0,0,0,0,0
xAccel,0,,,,,,,,
yAccel,0,,,,,,,,
zAccel,0.5,,,,,,,,
axAccel,-1,,,,,,,,
ayAccel,0,,,,,,,,
azAccel,0,,,,,,,,
fLMode,108,300,590,970,,,,,
fHMode,116,312,600,980,,,,,
fMin,65,,,,,,,,
fMax,1000,,,,,,,,
```

where each row has the following meaning:

• xHammer, yHammer and zHammer contain the location of the hammer in each test in cartesian co-ordinates. Each column corresponds to each hammer test in the order they appear in "C:\path\to\root\of\signal_calc\data". Note that the coordinate system you use is arbitray, but it must be consistent for both the impact and accelerometer locations.
• FxHammer, FyHammer and FzHammer contain the direction of each hammer impact.
• xAccel , yAccel and zAccel contain the locations of the accelerometers which are assumed to remain constant throughout.
• axAccel , ayAccel and azAccel contain the direciton of the accelerometers.
• fLMode and fHMode contain the upper and lower bounds of each mode in Hz. The modes are not automatically detected, so you must manually choose suitable ranges around each peak. You can typically get good enough estimtes from SignalCalc.
• fMin and fMax specify the miminmum and maximum frequency range to include when identifying the modes. The data outside this range is ignored.

3. Run the script

Call the top-level modal_analysis function to perform the modal analysis, usinf the following syntax:

modes = modal_analysis("C:\path\to\root\of\signal_calc\data")

which will process the data, generating many graphs showing the fit to the experimental data. The key output you will want is the modes structure, containing the following fields:

• modes.omega contains the natural frequencies in rad/s
• modes.zeta contains the damping ratios as a fraction
• modes.r contains the nodes of the model in cartesian co-ordinates. This will be a superset of your impact and accelerometer locations.
• modes.u contains the mass-normalised deflection at each node
• modes.n contains the direction of the deflection at each node in cartesian co-ordinates.

Example

A worked example has been included. This contains the results from a roving hammer test on a beam strucuture. An accelometer was placed at one location, and the beam was impacted a different locations along its length. Change your working directory to the root of ModeLAB and run the run_example.m script in the Example folder. This executes the following command to process the data and identify the modes:

modes = modal_analysis('.\Example\SignalCalc')

and then visualises the mode shapes of the beam as follows:

figure
for i = 1:4
    subplot(4,1,i)
    plot(modes.r(:,3), modes.u(:,i))
    title(sprintf('%0.2f Hz', modes.omega(i)/2/pi)
end

This should generate a plot like this: