Simcenter Testing Solutions Modal Assurance Criterion (MAC)

2019-08-29T16:35:15.000-0400

Simcenter Testlab

Summary

Details

Direct YouTube link: https://youtu.be/8j2xspUp6cY

The Modal Assurance Criterion or MAC calculation is a used in Modal Analysis to indicate the similarity of two mode shapes.

This article explains Modal Assurance Criterion (MAC). It has the following sections:
1. Background
2. Modal Assurance Criterion Equation
3. Modal Assurance Use Cases
4. Experimental Modal Analysis Application Example
5. FEA-Test Application Example
6. Calculating MAC in Simcenter Testlab
6.1 Auto-MAC
6.2 MAC Comparison
7. Calculating MAC in Simcenter Testlab Neo
8. Conclusion

1. Background

The Modal Assurance Criterion Analysis (MAC) analysis is used to determine the similarity of two mode shapes:

If the mode shapes are identical (i.e., all points move the same) the MAC will have a value of one or 100% as show in Figure 1.
If the mode shapes are very different, the MAC value will be close to zero, as shown in Figure 2.

If a mode shape was compared to itself, the Modal Assurance Criterion value should be one or 100%.

Figure 1: Mode shapes with 100% MAC value

For modes with different shapes, the MAC is less than 1. Shapes that are very different will have a value close to zero as shown in Figure 2.

Figure 2: Mode shapes with 1.5% MAC value

Mode shapes that are used in the comparison can originate from a Finite Element Analysis or from an experimental modal analysis.

In a typical MAC analysis, one might make a ‘MAC Matrix’. A ‘MAC Matrix’ is a series of bar graphs of MAC values, that each range from 0 to 100% as shown in Figure 3.

In the case of Figure 3, this is a mode set compared to itself. The mode set contains nine different individual modes, so 81 different MAC values are being calculated. About half the values are redundant –e.g., the MAC value between mode 1 and 3 is the same as between mode 3 and 1.

Figure 3: MAC matrix comparing a set of 9 modes. Red values are 100% MAC values. Dark blue is less than 10% MAC value.

In Figure 3, the first mode shape at 133 Hz is identical to itself, hence a single red bar of a value of 1. Along the diagonal, every mode is identical to itself, 1 to 1 (133 Hz), 2 to 2 (135 Hz), 3 to 3 (304 Hz), etc.

Off of the diagonal, the MAC values are very low. Ideally, each mode should be uniquely observed and have a different shape than the other modes. This is the case for this mode set. The highest off diagonal mode pair is mode 2 compared to mode 9 (and vice versa 9 to 2) with a MAC value of 20%. All the other off-diagonal mode pairs are below 20%.

2. Modal Assurance Criterion Equation

The MAC value between two modes is essentially the normalized dot product of the complex modal vector at each common nodes (i.e., points), as shown in Equation 1. It can also be thought of as the square of correlation between two modal vectors φr and φs.

Equation 1: Modal Assurance Criterion equation for comparing two mode shapes

If a linear relationship exists (i.e., the vectors move the same way) between the two complex vectors, the MAC value will be near to one. If they are linearly independent, the MAC value will be small (near zero).

A complex vector simply includes both amplitude and phase, whereas a real vector is real part only. In Equation 1, it is also clear that the MAC is not sensitive to scaling, so if all mode shape components are multiplied with the same factor, the MAC will not be affected.

If an experimental modal analysis had 20 different nodes where measurements were made, the mode shape components at all 20 nodes are taken into account to calculate the MAC value, but more importance will be attributed to the higher amplitude node locations.

3. Modal Assurance Use Cases

A Modal Assurance Criterion (or MAC) analysis can be used in several different ways:

FEA-Test comparison – A MAC can be used to compare modes from an experimental modal analysis test to a Finite Element Analysis (FEA) and an object as shown in Figure 4. It will indicate if the same mode shapes are found in both the test and FEA analysis.
FEA-FEA comparison – Several assumptions can be made in the creation of a FEA analysis: Young’s Modulus, boundary conditions, and mass density values to name a few. A MAC analysis can determine the degree to which these assumptions affect the resulting mode shapes.
Test-Test comparison – A MAC analysis can flag potential issues with the modal analysis results. Usually MAC will identify modes and areas that could benefit from acquiring more data points on the structure.

Figure 4: Experimental modal analysis (left) versus a finite element analysis (right) of an exhaust system

A MAC analysis is only looking at the mode shape, it does not compare the frequency value.

4. Experimental Modal Analysis Application Example

When performing an experimental modal analysis, the test operator must decide the number of points (i.e., nodes) to be measured. Determining the proper number is critical to the success of the test. If not enough points are measured, then the mode shape will not be identified properly.

Consider the case of an experimental modal analysis performed on a rectangular plate, suspended with free-free boundary conditions. Frequency Response Function (FRF) data was acquired at 6 locations on the plate. The FRF data was analyzed and a mode set extracted. After performing a MAC on the resulting mode set, not all the off-diagonal MAC values are close to zero (Figure 5).

Figure 5: For a 6 point experimental modal analysis of a plate, the off-diagonal modes are not all close to zero. Modes 3 (385 Hz) compared to mode 7 (764 Hz) has a MAC value of over 90% which is unusual.

Upon closer inspection of modes 3 and 7, the shapes themselves are unexpected (Figure 6). They appear to be rigid body modes. Rigid body modes on a free-free suspended structure are normally around 0 Hz. There are 6 of them: translation in X, Y, Z and rotation in X, Y, and Z.

At frequencies like 385 Hz and 764 Hz, flexible modes of the structure are expected. In this case, only measuring at 6 locations on the structure leads to “spatial aliasing”. There are not enough points to capture the modes correctly.

Figure 6: Both 385 Hz and 764 Hz appear to be rigid body modes, which is not expected.

Acquiring an additional 9 points leads to better results. With 15 total points, the mode shapes look completely different (Figure 7).

When viewing modes 3 and 7 again, but with 15 measured points, one can better appreciate how measuring only 6 points created the spatial aliasing error (Figure 6 versus Figure 7).

Figure 7: 15 point modal: 385 Hz on left, 766 Hz on right

When comparing Figure 6 (6 point modal) versus Figure 7 (15 point modal):

The 385 Hz mode was a bending mode along the center of the plate as seen in the 15 Point modal analysis. Because points were only acquired on the edges in the 6 point modal, the entire bending was missed, resulting in a mode shape that appeared to be a vertical translation rigid body mode.
The 766 Hz mode was actually a triple bending along the length of the plate as seen the 15 point modal analysis. The 6 point modal missed all the key bending areas, resulting in an apparent vertical translation rigid body mode.

The MAC analysis of the plate structure experimental modal analysis of the 15 point modal analysis is much improved as shown in Figure 8. The off-diagonal MAC values are much closer to zero.

Figure 8: MAC analysis with 6 measured points (left) versus 15 measured points (right)

In experimental modal analysis, the data measured in the 6 point modal analysis is not "wrong". The FRF measurements at these nodes were no different in the 6 point modal versus the 15 point, since the physical structure being tested did not change. There was simply not enough measurement points to determine the complete mode shape. This is different than a Finite Element modal analysis were the number of nodes does determine the dynamic behavior.

In this case, a Modal Assurance Criterion (MAC) analysis flagged a problem with an inadequate set of measurement points. Because the off-diagonal MAC values in the MAC matrix were not low, the error was easy to find.

5. FEA-Test Application Example

An experimental modal analysis was done on an exhaust system and compared to a finite element modal analysis of the same exhaust (Figure 9).

Figure 9: Experimental modal analysis (left) versus a finite element analysis (right) of an exhaust system

After collecting Frequency Response Functions (FRFs) on the exhaust system, a MAC analysis was done between the first thirteen experimental test modes and the first thirteen finite element analysis modes. The results are shown in Figure 10.

Figure 9: MAC Matrix comparing exhaust system finite element modes and experimental modes

Looking at the diagonal of the MAC matrix:

MAC values are not 100%, because the two sets of modes are not identical.
Modes 9 thru 13 are less than 75%.
Modes 11 and 13 are not in the same order between the two mode sets

In this case, the MAC analysis indicates that there is room for improvement in the correlation of the test and FEA.

Using a variant of the MAC analysis called a ‘MAC Contribution Analysis’, the nodal points which most reduced the average MAC below 100% can be identified. Performing this analysis, it was found that nodal points of the Y pipe part of the exhaust were most responsible for the decrease.

Using this information, a visual inspection of the actual exhaust found that welds were present at the joint of the Y pipe (Figure 10). These welds were not physically represented in the finite element model, which was created from a CAD geometry. This can happen because when weld locations are only indicated on CAD drawings, but not physically present.

Figure 10: Welds at the Y pipe joint of the physical exhaust system (left) had to be introduced into the finite element model (right) to achieve better correlation.

After introducing appropriate elements at the Y joint, the finite element model was recalculated and compared to the test results (Figure 11).

Figure 11: MAC analysis with updated FEA results versus test results

While not perfect, the following improvements were observed:

Lowest MAC value along the diagonal improved from 69% to 80%
Swapping of modes 11 and 13 was eliminated

Using the MAC and ‘MAC Contribution Analysis’ as guidance, the Finite Element model was significantly improved. This process can be continued until the results match within desired limits.

The most important part of this process is the actual “detective” work. Figuring out that welds need to be added to the FEA model (which are not in the base CAD parts) is an important lesson for future modelling projects. The MAC and ‘MAC Contribution Analysis’ are tools to be used in this process.

6. Calculating MAC in Simcenter Testlab

Direct YouTube link: https://youtu.be/qm02Z2Cdxe8

To calculate a Modal Assurance Criterion in Simcenter Testlab (formerly called LMS Test.Lab), use the ‘Modal Validation’ worksheet of Simcenter Testlab Modal Analysis (Figure 12).

Figure 12: To calculate MAC analysis, open the ‘Modal Validation’ worksheet

A MAC analysis can be performed:

‘MAC’ - Between two different mode sets
‘Auto - MAC’ - A mode set to itself

6.1 Auto-MAC

To calculate a MAC analysis within a single mode set, select the “Processing” set that contains the modes in the upper left of the ‘Modal Validation’ worksheet (Figure 13).

Figure 13: To calculate MAC of one mode set to itself, use “Auto-MAC”

Then press the “Auto-MAC” button in the middle left of the screen. After pressing the “Auto-MAC” button, a table of MAC values is created in the upper right.

The default view of the table is called “Table/Geometry”. Rather than looking at a table, a set of bars can be viewed by selecting “Matrix/Geometry” in the upper right (Figure 14).

Figure 14: Switch the MAC table view from the default “Table/Geometry” to “Matrix/Geometry”.

In the resulting Matrix view, one can click on the bars to see the MAC value for any mode pair. The corresponding mode shapes are automatically displayed below (Figure 15).

Figure 15: The MAC matrix view is a graphical representation of MAC values.

6.2 MAC Comparison

A MAC analysis can also be performed between two different modes sets. This is done by selecting two different “Processing” sets (Figure 16).

Figure 16: For comparison between two separate mode sets, used the “MAC” button after selecting the appropriate mode sets.

To do a comparison between two different mode sets, each needs to be identified in a separate location on the left side of the ‘Modal Validation’ worksheet:

First Mode Set – Select at the top via the “Active Processing” dropdown.
Second Mode Set – Select at the bottom with either the “Section” toggle or “Input Basket” toggle.

Press the “MAC” button to perform the analysis.

7. Calculating MAC in Simcenter Testlab Neo

Direct YouTube link: https://youtu.be/jPzM0vLzxtI

Modal Assurance Criterion calculations can be performed in Simcenter Testlab Neo with revision 2306 and higher. To do so, select "Modal Correlation" under "File -> Add-ins" in Simcenter Testlab Neo. It requires 25 tokens to run.

8. Conclusion

A Modal Assurance Criterion (or MAC) analysis can be used for FEA-Test, FEA-FEA and Test-Test comparisons of modes. By analyzing a MAC matrix, an engineer can improve the quality of an experimental modal test, verify finite element models, and update FEA models with test data.

Questions? Email peter.schaldenbrand@siemens.com, contact Siemens Support Center, or reply to this article.

Also, be sure to check out the Modal Testing Seminar.

Related Links:

Other Articles:

KB Article ID# KB000043133_EN_US

Summary Details

Associated Components

Simcenter Testlab Digital Image Correlation Testlab Environmental Testlab Acoustics Testlab Data Management Testlab Desktop Testlab Durability Testlab General Acquisition Testlab General Processing & Reporting Testlab Rotating Machinery & Engine Testlab Sound Designer Testlab Structural Dynamics Testlab Turbine

Simcenter Testing Solutions Modal Assurance Criterion (MAC)

Summary

Details

KB Article ID# KB000043133_EN_US

Contents

Associated Components