Modal test: geometry declaration and data acquisition/import

Contents Functions Index

PDF

2.2 Modal test: geometry declaration and data acquisition/import

Before actually taking measurements, it is good practice to prepare a wire frame-display (section 2.2.1 and section 4.1.4 for other examples) and define the sensor configuration (section 2.2.2).

The information is typically saved in a specific .m file which should look like the gartte demo without the various plot commands. The d_pre demo also talks about test preparation.

2.2.1 Modal test geometry declaration

A wire-frame model is composed of node and connectivity declarations.

Figure 2.5: Test analysis : wire-frame model.

Starting from scratch (if you have not imported your geometry from universal files). You can declare nodes and wire frame lines using the fecom Add editors. Test wire frames are simply groups of beam1 elements with an EGID set to -1. For example in the two bay truss (see section 4.1.4)

 cf=fecom;cf.model='reset';
 % fecom('AddNode') would open a dialog box
 fecom('AddNode',[0 1 0; 0 0 0]); % add nodes giving coordinates
 fecom('AddNode',[3  1 1 0;4  1 0 0]); % NodeId and xyz
 fecom('AddNode',[5      0 0 0    2 0 0;
                  6      0 0 0    2 1 0]);

 fecom('AddLine',[1 3 2 4 3]);  % continuous line in first group
 fecom('AddLine',[3 6 0 6 5 0 4 5 0 4 6]); % 0 for discontinuities
 fecom('ProInit');
 %fecom('save')  % will let you save the model to a mat file

Note that

After declaring nodes you can use cf.sel='groupall';fecom(cf,'AddLine') to start a cursor letting you build the wire-frame line graphically. Click on nodes continue the line, while the context menu allows breaks, last point removal, exit, and display of the ldraw building commands in the MATLAB command window. This procedure is particularly useful if you already have a FEM model of your test article.
you can also use the femesh ObjectBeamLine command to define the experimental mesh.
If you have a FE mesh, you can define the wireframe as a set of sensors, see section 3.1.1. An alternative is to combine the FE and wire frame models using the femesh AddTest command (see gartte). The procedure to solve the common problem of matching coordinate systems when they initially have different scales / orientations / origins is discussed in section 3.1.1.

The feplot and fecom functions provide a number of tools that are designed to help in visualizing test results. You should take the time to go through the gartid, gartte and gartco demos to learn more about them.

2.2.2 Sensor/shaker configurations

The SDT handles very general sensor configurations, see section 4.3 and section 4.3.3. For experimental modal analysis one however usually only considers translation and rotation sensors, which will be illustrated here.

Figure 2.6: Sensor/shaker locations.

Basic sensor configurations correspond to cases where the measurements are translations in global directions and, if test analysis correlation is desired, the sensors are located at finite element nodes. In such cases, everything can be easily handled using DOF definition vectors.

DOF definition vectors (see mdof) allow the description of translation DOFs in global directions. The convention that DOFs .07 to .09 correspond to translations in the −x,−y,−z directions is implemented specifically for the common case where test sensors are oriented this way. For example, you can display sensors using (see the gartte demo). Note that this example shows 3 different ways to define the translation sensors: from DOF, from DOF with a local basis, and with the tdof format ([SensID NodeID tx ty tz] giving a sensor identifier (integer or real), a node identifier (positive integer), and the projection of the measurement direction on the global axes).

cf=demosdt('demogartteplot')

% simply give DOFs
cf.mdl=fe_case(cf.mdl,'sensdof','Test', ...
[1011.03 1001.03 2012.07 1012.03 2005.07 1005.03 1008.03 ...
1111.03 1101.03 2112.07 1112.03 2105.07 1105.03 1108.03 1201.07]');

% Give DOF defined in a local basis
cf.mdl=fe_case(cf.mdl,'sensdof append','Test', ...
[2201.01 1; 3201.03 0; 1206.03 0; 1205.01 1; 1302.01 1]);

% Give identifier, node and measurement direction
cf.mdl=fe_case(cf.mdl,'sensdof append','Test', ...
[1 2301 -1 0 0; 2 1301 0 0 1; 3 2303 -1 0 0; 4 1303 0 0 1]);
fecom('curtab Cases','Test');

% You can now display FRFs using
ci=iicom('curveload gartid');idcom('e .05 6.5')
cf.def=ci.Stack{'IdAlt'}; % automatically uses sensor definition 'Test'

Sensor and shaker definitions associated with each measured input/output pair are normally entered in the acquisition phase as detailed in section 2.1.4. Except for roving hammer tests, the number of input locations is usually small and only used for MIMO identification (see section 2.4).

Once a sensor configuration defined, you can directly animate measured shapes (called Operational Deflection Shapes) as detailed in section 2.2.4.

For test/analysis correlation, see section 3.1. For more advanced sensor definitions see section 4.3. For interpolation of unmeasured DOFs see section 3.3.2.

2.2.3 Data acquisition

The Structural Dynamics Toolbox does not intend to support the acquisition of test data since tight integration of acquisition hardware and software is mandatory. A number of signal processing tools is gradually being introduced in iiplot (see ii_mmif FFT or fe_curve h1h2). But the current intent is not to use SDT as an acquisition driver. The following example generates transfers from time domain data

frame=fe_curve('testacq');  % 3 DOF system response
% Time vector in .X field, measurements in .Y columns
frf=fe_curve('h1h2 1',frame); % compute FRF
[ci,XF]=iicom('curveid');XF('Test').w=frf.X; XF('Test').xf=frf.H1;
iicom('sub');

You can find theoretical information on data acquisition for modal analysis in Refs. [2][3][4][5][6].

Import procedures are described in section 2.1.4. The following table gives a partial list of systems with which the SDT has been successfully interfaced.

Vendor	Procedure used
Bruel & Kjaer	Export data from Pulse to the UFF and read into SDT with ufread or use the Bridge To Matlab software and pulse2sdt.
Dactron	Export data from RT-Pro software to the UFF. Use the Active-X API to drive the Photon from MATLAB see photon.
LMS	Export data from LMS CADA-X to UFF.
MathWorks	Use Data Acquisition and Signal Processing toolboxes to estimate FRFs and create a script to fill in SDT information (see section 2.1.4).
MTS	Export data from IDEAS-Pro software to UFF.
Polytec	Export data from PSV software to UFF.
Spectral Dynamics	Create a Matlab script to format data from SigLab to SDT format.

2.2.4 Operational deflection shapes

Operational Deflection Shapes is a generic name used to designate the spatial relation of forced vibration measured at two or more sensors. Time responses of simultaneously acquired measurements, frequency responses to a possibly unknown input, transfer functions, transmissibilities, ... are example of ODS.

When displaying responses with iiplot and a test geometry with feplot, iiplot supports an ODS cursor. Run demosdt('DemoGartteOds') then open the context menu associated with any iiplot axis and select ODS Cursor. The deflection show in the feplot figure will change as you move the cursor in the iiplot window.

More generally, you can use fecom Initdef commands to display any shape as soon as you have a defined geometry and a response at DOFs. The Deformations tab of the feplot properties figure then lets you select deformations within a set.

 [cf,XF]=demosdt('DemoGartteOds')
 cf.def=XF(1);
 % or the low level call : cf.def={XF(1).xf,XF(1).dof,XF(1).w}
 fecom('curtab Plot');

You can also display the actual measurements as arrows using

 [cf,XF]=demosdt('DemoGartteOds'); cf.def=XF(1);
 cf.sens=XF(1).dof;fecom showarrow;

For a tutorial on the use of feplot see section 4.1.