Modal Testing: Driving Point Survey – What’s at stake?

smacdon
Siemens Experimenter
08-29-2018 09:27 PM
(view in My Videos)

A driving point survey is an important part of any modal analysis test.  By making several
driving point measurements at various points around our structure we can “survey” the structure
to help determine the optimal location for excitation.  Consider the three driving point FRFs in
Figure 1 below.

Figure 1: Three different driving point FRFs from the same structure.

Which FRF in Figure 1 contains the most information about the natural frequencies of the
structure?  It is clear the blue FRF shows three peaks, while the red and the green FRFs each
only show two peaks.  By looking at the set of FRFs and comparing them, we can conclude that
the location used to measure the blue FRF is the best driving point of the three, because the blue
FRF contains the most peaks.  This example highlights why performing a driving point survey is
an important step in a modal test!

This article will explain:

 What a driving point measurement is

What is a “driving point” measurement?

A driving point measurement is a dynamic measurement where the force input from the hammer
(or shaker) and the response output from the accelerometer are measured at the same point on the
structure, and in the same direction.  (See Figure 2)

Figure 2: A driving point measurement on a cantilever beam. Both the force and response are
being measured at the same location on the structure, and in the same direction (-Z).

However, sometimes it may not be physically possible to collocate the impact and the response
measurement as shown in Figure 2, and a compromise must be made.  For instance, imagine
making a driving point measurement on a fuel tank, or other sealed vessel, where the interior
surface is not accessible (see Figure 3).  In this case, the accelerometer needs to be mounted on
the same surface we will impact with the hammer, and should be located as close as possible to
the impact location.  It is important that the accelerometer be located where it will not interfere
with the impact itself.  Under no circumstances should the impact hammer be striking the
accelerometer, or making any direct contact, as this will distort the FRF we are trying to measure
(and likely overload the signal coming from the accel).

Figure
3 - When necessary, place accelerometer next to, and as close as possible, to the impact location.

While it is critical that the force and the response be measured at the same location and direction,
the polarity of the two measurement directions can be different.  For example, the hammer can
be measuring in the -Z direction, while the accelerometer is measuring in the +Z direction.  This
is often the case in situations where the accel is mounted next to the impact location like is
shown in Figure 3.  As long as both hammer and accel signals are measuring along the same
axis, the driving point measurement is valid.

Why is a driving point measurement important?

Very simply - a driving point FRF shows all of the modes of a structure that are excited by
impacting at a particular location.  If impacting at a point on a structure does not adequately
excite a particular mode shape, the peak corresponding to that mode will be missing from the
FRF.  Why? To find out, let’s take a look at how the FRFs from Figure 1 were generated. 

Consider the cantilever beam in Figure 4a below.  By placing the accelerometer at point R and
impacting at that same location, we get the red FRF shown in Figure 4b.  Next, we move the
accelerometer and impact at point G, which generates the green FRF.  Lastly, we move the
accelerometer and impact to location B, and measure the blue FRF.  Again, we notice that there
are three peaks in the blue FRF, and only two in the red and green FRFs.

Figure 4: (a) Cantilever beam with three driving points –“R”, “G”, and “B”; (b) driving point
FRFs from each location
To understand why this is, let’s take a look at the first three modes of vibration for our cantilever
beam in Figure 5 below. 

Figure 5: The first three mode shapes of a cantilever beam. The higher mode shapes feature
points that do not move called nodes.

In the first mode, the entire beam participates in the deflection pattern, all points are moving.  In
the 2nd and 3rd mode however, there are points that do not move at all.  These points that do not
participate (ie- move) in the deflection shape are called nodes, or nodal points.

Overlaying our cantilever beam with the first three mode shapes (Figure 6), it becomes clear that
the Red impact location is on top of a node in the 3rd mode shape, and the Green impact location
is at a node of the 2nd mode. 

Figure 6: Nodal points in the 2nd and 3rd modes occur at Green and Red driving points
respectively. These modes will not be excited by impacting at these locations!

Because nodal points do not participate in mode shapes, impacting our structure at the Red and
Green locations will not adequately excite the mode shapes that have nodes at these locations. 
As a result, the peaks corresponding to the missing modes will not appear in the FRF.    

This is the most important function of a driving point measurement: A driving point FRF shows
us which mode shapes are being adequately excited by impacting our structure at that particular
location.  The modes that are excited by impacting at that location will create peaks in our
driving point FRF, the modes that are not excited by that impact location will not. 

Driving Point Survey

By making several driving point measurements at various locations around the structure and
comparing the FRFs, we can tell which location will best excite the modes we are interested in. 
This is known as a “driving point survey.” 

Performing a driving point survey helps to avoid using a nodal point for excitation.  For a
realistic structure with 2 or 3 critical dimensions, this becomes even more important because the
nodal “points” from the simple cantilever beam example turn into nodal “lines” as shown in
Figure 7 below.

 
Figure 7: Mode of a circular disc featuring two nodal lines. Points along these lines do not
participate in the mode shape.

However, by performing a driving point survey we gather the information needed to avoid nodal
lines, and ensure the modes of interest will be properly excited and measured.

Driving point measurements are important for other reasons as well.  As we learned in
this Modal Tips Article, in order to view mode shapes, it is necessary to measure either a
complete row or complete column of our measurement matrix.  The driving point measurements
represent the diagonal of the test matrix, and will always be part of this comprehensive modal
survey data set.   Driving point measurements also allow us to properly scale mode shapes, and
calculate modal mass and stiffness for a structure.  Unless we are only interested in resonant
frequencies of a structure, a driving point measurement is always required.

Using Simcenter Testlab Impact Setup: Driving Points

After completing the setup of the trigger, bandwidth, and windowing using the steps in the
Impact Setup workflow, the next step is the driving point survey.

One tip that makes the Driving Points process a little simpler is to move the Geometry tab in
front of the Channel Setup tab in the Testlab workflow as shown in Figure 8.  If you’d like to
learn how to customize the workflow in Testlab, please see this article in the Simcenter Testing
Knowledge Base.

Figure 8: Move Geometry ahead of Channel Setup tab to make the Driving Point survey easier.

Setting up our test structure in Geometry is a good thing to do before we begin acquiring data, as
we can use it to name the input channels and ensure our data Point IDs match our geometry Point
IDs.  This is one of the reasons moving Geometry before Channel Setup can be helpful. 

For this example the test structure is a flat rectangular plate, made up of 15 nodes, as shown in
Figure 9 below.

 
Figure 9: Test Geometry of a flat plate.

For this example let’s say we need to know the first 8 modes of this plate, their natural
frequencies and mode shape description.  Since I am unfamiliar with this plate, I select several
points around the structure at which to make driving point measurements.  The selected points
are plate:2, plate:4, plate:7, plate:8, and plate:15, as shown below in Figure 10.

 
Figure 10: Point IDs selected to act as driving points.

Now that our test geometry is done and our driving point locations are identified, we can move to
the Channel Setup tab and begin the driving point survey.  At this point we should have 2
channels turned on – a Force channel (Input 1) and an Acceleration channel (Input 2).  Channel
Setup is shown in Figure 11 below.

Figure 11: Channel Setup for our impact test.

 
Next we will incorporate our geometry into Channel Setup by clicking on the down arrow next to
“Channel Setup” in the blue bar near the top of the window as shown in Figure 12, and click on
“Use Geometry”.

Figure 12: Select “Use Geometry” to

This will open up a new window pane in Channel Setup for our geometry.  Click on “Refresh” to
show the structure (Figure 13).

Figure 13: “Use Geometry” in Channel Setup to aid in naming input channels.

 
Our first driving point will be at the point named “plate:2” as shown in Figure 14 below.  I can
select this point in the geometry by clicking on the node icon or by highlighting it in the point ID
list by clicking on the row header.  Next, select the destination channels on the left half of the
screen, again by clicking on the row headers for Input1 & Input 2 while holding the [SHIFT]
key.  Clicking “Insert” at the bottom of the screen will move the selected Point ID (plate:2) over
to our two channel Point IDs (Figure 15).

Figure 14: Select the first test point in the geometry and insert the ID on input channels.
Figure 15: Channels 1 & 2 renamed to “plate:2” to match the geometry point ID!

With the measurement channels correctly labelled for the first driving point (plate:2), we can
move to the Impact Setup tab in the Testlab workflow, and into the Driving Points page (Figure
16).

Figure 16: Driving

The Driving Points worksheet contains multiple displays, highlighting different aspects of our
driving point measurement.  These areas are listed below:

A:  Time history (upper) and Autopower PSD (lower) of force channel

B:  Time history (upper) and Autopower PSD (lower) of response channel

C:  Instantaneous (current) driving point FRF Magnitude (upper) and Phase (lower)

D:  Average of driving point FRFs Magnitude (upper) and Phase (lower)

E:  Coherence

F:  Control & Setup area for measurement

If we look in the area labeled “F” in Figure 17 we see that the “Input point” is already filled in as
“plate:2”.  This is because we selected it in Channel Setup.  The “Response channel” is set to “2”
to indicate that our accelerometer is plugged into Channel 2 on our frontend.  We can select the
number of averages we’d like to use for our driving point FRF, as well as whether we want to
implicitly accept each average, or if we’d like to explicitly accept (via a popup window) after
each average.  These settings are shown in detail in Figure 18 below.  Click on “Start” to acquire
the driving point measurements for the first point. 

 
 Figure 18:
Input point will automatically read the Point ID from Channel Setup for the driving point
measurement.

After we have performed the 3 averages, the driving point measurement name will appear in the
“Driving Points” box in area “F”, and we are ready to move the accelerometer to our next
location, “plate:4”.  To update the channel information, we repeat the steps shown in Figure 14:
first select the node in the geometry, then insert the name into channels 1 & 2.  Once the channel
setup info is updated, return to the Driving Points worksheet, and it will appear as shown in
Figure 19. 

 
 Figure 19:
Use Channel Setup to update the naming of the next driving point measurement.

 
Repeat this process for the rest of the planned driving points.  When finished there will be a total
of 5 FRFs (Figure 20).   By selecting them all and clicking on “Display” we will see all 5 curves
overlaid (see Figure 21 below).

Figure 20: Select all driving point FRFs and

Figure 21: Driving point measurements overlaid

 
The driving point measurements will also appear in our Project, in a folder called “Driving
Points”, and can be plotted in Navigator.  By comparing all five driving point FRFs we can
quickly see some big differences between the excitation locations (Figure 22). 

Figure 22: All 5 driving point FRFs in a multi-trace display. Some driving point locations did not
excite all modes!

Each driving point location features a different number of peaks in the FRF.  This indicates that
only certain modes are being excited by impacting our structure at those locations.  By making a
table of the resonant peaks in each FRF, we can more easily see which driving point location will
work best for this structure (Figure 23). 

 
Figure 23: Driving point location comparison. Only plate:15 excited all 8 modes of interest.

The red boxes in Figure 23 indicate that there is no peak at that frequency in the FRF (orange
indicates weak excitation).  Once we count up the modes and resonant frequencies found in each
of the 5 driving point FRFs, we quickly see that only the FRF measured at plate:15 excited all 8
modes of interest.  Every other impact location missed at least 2 modes we are interested in.  By
looking at the mode shape descriptions in Figure 23 for each of the modes, it becomes clearer
why certain modes are not excited by impacts at certain locations.  For example, consider the
torsion mode, Mode #2 in Figure 23.  Only points that are not on the two center lines of the plate
will participate in this mode shape.  In this case, only plate:4 & plate:15 are not on a center-line. 
(See Figure 24)

 
Figure 24: Points that are on a center-line of the plate will not participate in the torsion mode.

The first 8 mode shapes of the plate are shown below in Figures 25-32. 

 
Figure 25: Mode 1 - Bending
Figure 26: Mode 2 - Torsion
Figure 27: Mode 3 - 2nd Torsion
Figure 28: Mode 4 - 2nd Bending
Figure 29: Mode 5 - Longitudinal Bending
Figure 30: Mode 6 - 3rd Torsion
Figure 31: Mode 7 - 2nd Longitudinal Torsion
Figure 32: Mode 8 - 3rd Bending  

In general it is best to avoid centerlines of our structure, focusing on corners and impact
locations at the extreme ends and edges of the structure.  This will typically excite the most
modes and avoid nodal lines of the first few modes, which are generally of primary interest. 
However, on a new or unfamiliar structure, it is always in the best interest of the tester to
perform a driving point survey.  This will help to ensure that the structure is adequately excited,
and the critical mode shapes of interest are measured.

Questions?  Email Scott MacDonald ([email protected]) or contact GTAC support.

Related Articles:
   Natural Frequency and Resonance
 Testlab Impact Testing
 Modal Tips: Roving Hammer vs. Roving Accel
 Modal Tips: What Tip Should I Use for My Impact Hammer?
 Testlab Impact Testing: User Defined Impact Sequence
 How to calculate damping from an FRF?
 What is a Frequency Response Function (FRF)?
 Nastran and Test: Compare Mode Shapes and FRFs
 Simcenter Testlab Modal Analysis: Modification Prediction
 Modal Impact Testing: User Defined Impact Sequence
 Import CAD into Simcenter Testlab
 Animate CAD Geometry
 Alias Table: Mapping Test Data to Geometry
 Geometry in Simcenter Testlab
 Maximum Likelihood estimation of a Modal Model (MLMM)
 Ground Vibration Testing and Flutter

Labels:

 Simcenter Testlab
 Structural Dynamics Testing

Share
Comment
Comments
DungLee
Experimenter
a week ago

 First of all, Thank you for your post.

But, i want to have some questions.

1. Why driving point measurements need to do at the same DOF (same point and direction, for
example)?

For example, i mount accel at DOF 2 and apply Force at DOF 1 instead of both accel and Force
mount at DOF 1 (or 2).

2. Does we need to choise how many driving point measurements are proper (3 , 5 or more...)
What amount of driving points depend on?

 
Thank you so much!

Dung LE

smacdon
Siemens Experimenter
a week ago

Hi @DungLee - Glad you enjoyed the article.  Hopefully I can answer your questions, and
provide additional insight.

1.  The reason we do a "driving point" measurement is to determine a good place to excite the
structure.  By placing the accelerometer and the force input at the same DOF and direction we
are measuring the effectiveness of that particular DOF as an excitation location.  The FRF from a
driving point will show us ALL of the modes we are exciting by impacting the structure at that
particular DOF. This is where the name "driving point" comes from:  the response is measured at
the "driving" or "forcing" location.  You could think of the driving point as being a "forcing
point" FRF.

Once the accel and the force input are at different locations (like in your question, accel at DOF
2 and force at DOF 1) we are no longer measuring a "driving point" FRF.  We are measuring a
FRF between those two locations.  (ie - response/input = DOF2/DOF1)  This FRF is certainly
important for understanding the structure, and for viewing mode shapes, but once the accel and
the hammer are in different locations (or directions), it is by definition no longer a "driving
point" FRF.  

2.  We need to try a couple different driving points to be able to compare the FRFs and the
number of peaks.  So, a minimum of two driving point FRFs are required in order to compare the
two, and see which excites more modes.  At a maximum, you could perform a driving point FRF
at every DOF/point in your geometry, and compare all of the FRFs.  Depending on your
structure's symmetry and layout, this strategy may produce redundant information, but you
would know for sure which driving point is best at exciting the most modes of your structure.  As
you gain more experience with a particular structure you will probably only need to compare 3-4
forcing locations to find the best one.

DungLee
Experimenter
Thursday

Hello, Smacdon
First, thank you for your meticulously reply.

for question 2. I understood.

But for question 1, some problems:

1. according to definition of driving point, i understood that: any measurement that is done with
both accel and force are mounted at the same DOF is called "driving point measurement". is it
right or wrong?

2. If right, in example for the cantilever beam above, you (or anyone) did 3 driving point
measurements (point R, G and B). And, as you wrote: "The FRF from a driving point will show
us ALL of the modes we are exciting by impacting the structure at that particular DOF". But, let
see at point R FRF (or point G FRF). It shows us two peaks but not three peaks. 

3. Assume that, "The FRF from a driving point will show us ALL of the modes we are exciting
by impacting the structure at that particular DOF" is true. So, Why does the FRF from a driving
point will show us ALL of the modes but the FRF from a non - driving point is not?

Thank you so much !

and please forgive me for anything

smacdon
Siemens Experimenter
Thursday

Hi @DungLee - You are right on in #1.  Any FRF measurement where the force and acceleration
are measured at the same location and direction (DOF) is a driving point FRF.  

#2)  Yes, the driving point FRFs from location R & G only show two peaks, while driving point
from B shows three.  This is exactly *why* we are doing multiple driving point FRFs!  We need
to compare all three driving point FRFs and count the number of peaks in each.  ONLY point B
is a suitable impact location if we are looking for the first three modes, because it is the only
DOF we tried that excites all three modes.  By hitting the cantilever beam at point R or G, we are
hitting on a nodal point.  Nodes don't participate in mode shapes, so putting force into the
structure there doesn't work - that mode shape will not be excited. 

 
#3)  The reason a non-driving point FRF doesn't *necessarily* show us all the modes excited by
our impact DOF is because not all DOF participate in a particular mode shape.  So it will depend
on which point you have placed the accel.  From the plate example in the article - consider point
8, in the center of the plate.  This point does NOT participate/move in several of the modeshapes
(mode 2, 3, 4, 6, or 7).  So, if I impact at DOF #15 (the best impact location from my survey),
place my accel on DOF #8 and look at the FRF (DOF8/DOF15) I will not see peaks at the
natural frequencies associated with modes 2, 3, 4, 6, or 7.  This is because even though impacting
at DOF15 excites all the modes I'm interested in, DOF 8 doesn't participate in all of them.  This
is why we only use driving point FRFs where the impact and accel are at the same DOF to
determine whether a particular impact DOF is suitable for our needs.  If the mode is excited by
impacting at that DOF, the accel WILL participate, and we will have a peak in our FRF for that
mode.

Hope this helps!

DungLee
Experimenter
Friday

Yes. Thank you so much.

I understand now.