App Note 56 www.vibetech.

com 6/1/2021

MEscope Application Note 56

Using MAC & SDI to Detect Faults in a Rotating Machine
The steps in this Application Note can be carried out using any MEscope Package that includes the VES-3600
Advanced Signal Processing & VES-4000 Modal Analysis options. Without these options, you can still carry out
the steps in this App Note using the AppNote56 project file. These steps might also require MEscope software with
a more recent release date.

APP NOTE 56 PROJECT FILE

• To retrieve the Project for this App Note, click here to download AppNote56.zip
This Project file contains numbered Hotkeys & Scripts for carrying out the steps of this App Note.
• Hold down the Ctrl key and click on a Hotkey to display its Script window
INTRODUCTION
MEscope contains two different metrics for comparing shapes, called the Modal Assurance Criterion (MAC) and
the Shape Difference Indicator (SDI). These two metrics are used to numerically compare a pair shapes, whether
they are Operating Deflection Shapes (ODS’s), mode shapes, or any shape data in a Shape Table (SHP file). Both
metrics have the same range of values between 0 & 1, but they have different properties.

MAC is a measure of the co-linearity of two shape vectors.

If two shapes lie on the same straight line, they are said to be co-linear and their MAC value equals 1.0.
If two shapes don’t lie on the same straight line, they are said to be linearly independent of one another, and their
MAC value is less than 1.0.

In MEscope, the following rules of thumb are used with MAC

MAC values ➔ between 0 &1
MAC = 1.0 ➔ two shapes are co-linear
MAC >= 0.9 ➔ two shapes are similar
MAC < 0.9 ➔ two shapes are different (or linearly independent)

SDI is a measure of the equality of two shape vectors.

If two shapes have equal components, their SDI value equals 1.0.
If two shapes do not have equal components, their SDI value is less than 1.0.

In MEscope, the following rules of thumb are used with SDI

SDI values ➔ between 0 &1
SDI = 1.0 ➔ two shapes have equal components
SDI >= 0.9 ➔ two shapes have similar components
SDI < 0.9 ➔ two shapes are different (some components are not equal)

Page 1 of 18
App Note 56 www.vibetech.com 6/1/2021

Several different commands in MEscope utilize both MAC and SDI.

STRUCTURE (STR) WINDOW COMMANDS

Animate a Pair | Animate Shapes: If Show MAC or Show SDI is checked on the Animation tab in the File |
Structure Options box, MAC or SDI, or both are displayed when two shapes are displayed side-by-side in
animation.

Comparing the First Order ODS of an Unbalanced Versus a Balanced Machine.

Page 2 of 18
App Note 56 www.vibetech.com 6/1/2021

SHAPE TABLE (SHP) WINDOW COMMANDS

Display | MAC: Displays a bar chart or spreadsheet of MAC values between pairs of shapes in two Shape Tables.

MAC Between Pairs of FEA & EMA Mode Shapes

Display | SDI: Displays a bar chart or spreadsheet of SDI values between pairs of shapes in two Shape Tables.

SDI Between pairs of FEA & EMA Mode Shapes

Page 3 of 18
App Note 56 www.vibetech.com 6/1/2021

Tools | Scaling | Residues to UMM Shapes: Displays a selection box where references of multi-reference
Residue mode shapes are selected for conversion to UMM mode shapes.

Display Box for Selecting References From Residue Mode Shapes.

DATA BLOCK (BLK) WINDOW COMMANDS

Tools | Data Block Correlation: Creates a new Data Block of MAC & SDI values for all samples of M#s with
matching DOFs in two Data Blocks.

Comparison of Two Data Blocks Using Tools | Data Block Correlation.

Page 4 of 18
App Note 56 www.vibetech.com 6/1/2021

Tools | M# Pairs Correlation: Creates a Shape Table of MAC & SDI values for all M#s with matching DOFs in
two Data Blocks.

Comparison of Two Data Block Using Tools | M# Pairs Correlation.

Frequency Response Assurance Criterion (FRAC): FRAC is displayed during FRF-based curve fitting in
MEscope. FRAC is the MAC value between each M# and its red Fit Function

FRAC Between an FRF & its Fit Function During FRF-Based Curve Fitting.

MODAL ASSURANCE CRITERION (MAC)

MAC is used to numerically compare a pair of complex valued vectors. This metric was the topic of a PhD
dissertation, and was originally used for comparing mode shapes in experimental modal analysis, hence its name [1].
[1] - R. J. Allemang "The Modal Assurance Criterion (MAC): Twenty Years of Use and Abuse", Proceedings of the
International Modal Analysis Conference, 2002
MAC measures the co-linearity of two complex vectors. Given two complex vectors {u} & {v} with matching
DOFs, MAC is calculated using the formula,
𝟐
‖{𝐮}𝐡 {𝐯}‖
𝐌𝐀𝐂 =
{𝐮}𝐡 {𝐮} {𝐯}𝐡 {𝐯}

Page 5 of 18
App Note 56 www.vibetech.com 6/1/2021

{u} is an n-vector
{v} is an n-vector
n ➔ number of vector components with matching DOFs (n >= 1)
h ➔ denotes the transposed complex conjugate (Hermitian) vector
‖ ‖𝟐 ➔ denotes the magnitude squared
MAC values ➔ between 0 &1
MAC = 1.0 ➔ two vectors are co-linear
MAC > 0.9 ➔ two vectors are similar
MAC < 0.9 ➔ two vectors are different (or linearly independent)

MAC PLOT
MAC is plotted below between two shape vectors {A} = {1, 1} and {B} = {x, y} where x & y have values between -3
to +3. MAC has a value of “1” anywhere along the straight line that passes through {0, 0} & {1, 1}. Those are the
values where the two vectors {A} & {B} are co-linear. For all other values of {B}, MAC has values less than “1”
but greater than or equal to “0”.

For values of {A} & {B} close to the origin (0, 0), MAC decreases more rapidly to zero when {A} & {B} are not
co-linear.
MAC is a more sensitive indicator when {A} & {B} have components with small numbers in them.

MAC for {A} = {1. 1} & {B} = {x. y} x = -3 to +3, y = -3 to +3.

MAC BETWEEN TWO NUMBERS

When n=1, meaning that {A} & {B} are scalars, MAC always has a value of “1”.
To obtain meaningful results, MAC must be applied to vectors with at least two components in them.

‖{𝐮}∗ {𝐯}‖𝟐 {𝐮}∗ {𝐮} {𝐯}∗ {𝐯}

𝐌𝐀𝐂 = = =𝟏
{𝐮}∗ {𝐮} {𝐯}∗ {𝐯} {𝐮}∗ {𝐮} {𝐯}∗ {𝐯}
n=1
* ➔ denotes the complex conjugate

Page 6 of 18
App Note 56 www.vibetech.com 6/1/2021

This property is the same as the Coherence function, which is used to measure linear dependence between two
DOFs of a cross-channel measurement function such as an FRF.

SHAPE DIFFERENCE INDICATOR (SDI)

SDI is also used to numerically compare a pair of complex valued vectors. This metric was created by the engineers
at Vibrant Technology as a complement to MAC.
Given two complex vectors {u} & {v} with matching DOFs, SDI is calculated using the formula,
𝟐 𝟐
‖{𝐮} − {𝐯}‖𝟐 𝟐 𝐑𝐞({𝐮}𝐡 {𝐯})
𝐒𝐃𝐈 = (𝟏 − ) =( 𝐡 )
{𝐮}𝐡 {𝐮} + {𝐯}𝐡 {𝐯} {𝐮} {𝐮} + {𝐯}𝐡 {𝐯}
{u} is an n-vector
{v} is an n-vector
n ➔ number of shape components with matching DOFs (n >= 1)
h ➔ denotes the transposed complex conjugate (Hermitian) vector
‖ ‖2 ➔ denotes magnitude squared
Re ( ) ➔ denotes the Real Part
SDI values ➔ between 0 &1
SDI = 1.0 ➔ two shapes have equal components
SDI > 0.9 ➔ two shapes are similar components
SDI < 0.9 ➔ two shapes are different (some components are not equal)

SDI BETWEEN TWO NUMBERS

When (n=1), the equation for SDI is reduced to a simpler formula,
𝟐
𝟐 𝐑𝐞({𝐮}∗ {𝐯})
𝐒𝐃𝐈 = ( ∗ )
{𝐮} {𝐮} + {𝐯}∗ {𝐯}
* ➔ denotes complex conjugate

SDI can be used to compare two scalars.

Page 7 of 18
App Note 56 www.vibetech.com 6/1/2021

SDI PLOT
SDI is plotted below between two shape vectors {A} = {1, 1} and {B} = {x, y} where x & y have values for -5 to +5.
• When the two vectors {A} & {B} have equal components, SDI = 1, but one vector can be the negative of
the other

For values of {A} & {B} close to the origin (0, 0), SDI decreases more rapidly to zero.
SDI is a more sensitive indicator when {A} & {B} have components with small numbers in them.

SDI for {A} = {1. 1} & {B} = {x. y} x = -5 to +5, y = -5 to +5.

ROUTE-BASED MACHINE MONITORING

In a route-based machinery health monitoring program, accelerometers are typically attached with magnets to a
machine and data is acquired with a portable multi-channel analyzer. On a motor-pump configuration, two
accelerometers are typically attached on the motor and two on the pump.
A 4-channel portable analyzer, a tri-axial accelerometer, and a uni-axial accelerometer are typically used to acquire
data while the machine is running.

During data acquisition, the uni-axial accelerometer remains fixed as a reference, and the tri-axial accelerometer is
moved to each test point prior to another acquisition.

Page 8 of 18
App Note 56 www.vibetech.com 6/1/2021

4-Channel Acquisition Using a Roving Tri-Axial & Fixed Reference Uni-Axial Accelerometer.
The rotating machine shown above was used to simulate route-based data acquisition. Data was acquired while the
machine was running at 2000 RPM. The tri-axial accelerometer was moved to four different locations between
acquisitions, two on the motor and one on each rotor bearing block.
Three ODS-FRFs were calculated for each test point, between the three responses measured by the tri-axial
accelerometer and the response of the uni-axial accelerometer.
An ODS-FRF is a cross-channel measurement. Its magnitude is the Auto spectrum of the Roving response and its
phase is the phase of the cross spectrum between the roving & reference responses. Because its magnitude is the
Auto spectrum of the Roving response, an ODS-FRF can be integrated from acceleration to velocity units.
The first-order ODS of the rotating machine was derived from the first-order peaks in a set of 12 ODS-FRFs.

The RPM of the machine can change slightly due to warmup of the motor and bearings causing the order peaks to
shift slightly in frequency. Therefore, a Peak cursor is used to obtain the first-order ODS of the machine.

Page 9 of 18
App Note 56 www.vibetech.com 6/1/2021

The project for this App Note contains nine different Data Block (BLK) files with 12 ODS-FRFs in each Data
Block, as shown below.
• Each BLK file contains 12 ODS-FRFs, describing the 3D vibration of the machine at four points
• Four sets of ODS-FRFs were calculated from data acquired while the two rotors were balanced
• One bad measurement was made by rotating the tri-axial accelerometer 90 degrees at Point 1
• Four sets of ODS-FRFs were acquired while unbalance screws were inserted in the rotors
• A Flat Top window was applied to each time waveform to improve the accuracy of the order-based
ODS’s
• A Flat Top window also widens the order peaks

Data Blocks of ODS-FRFs Acquired from the Rotating Machine.

Page 10 of 18
App Note 56 www.vibetech.com 6/1/2021

THREE USES OF MAC & SDI

In this App Note, MAC & SDI are used in three different ways,
1. To validate repetitive ODS-FRF measurements from the same rotating machine
2. To locate a bad ODS-FRF measurement
3. To diagnose several machine unbalance cases

Unbalance was Created by Adding Screws to the Inboard & Outboard Rotors.

STEP 1 - EXPORT DATA BLOCKS OF ODS-FRFs

In this step, nine Data Blocks, each containing 12 ODS-FRFs, are exported from MEscope so that they can be
imported, one at a time.
• Press Hotkey 1 Export ODS-FRFs

The Data Blocks are saved in the same folder as the MEscope program folder.

STEP 2 - IMPORT A DATA BLOCK OF ODS-FRFs

• Press Hotkey 2 Import ODS-FRFs
• Select Balanced Case-1.BLK in the dialog box shown below

Page 11 of 18
App Note 56 www.vibetech.com 6/1/2021

The first-order ODS of the balanced machine at its running speed (about 2000 RPM) is displayed in animation
on the machine model as shown below.

First-order ODS of the Balanced Machine.

Page 12 of 18
App Note 56 www.vibetech.com 6/1/2021

STEP 3 - [MAG & PHS] OF THE ODS

• Press Hotkey 3 ODS [Mag & Phs]

[Mag & Phs]of the First-order ODS of the Balanced Machine.

[Mag+Phs] flags are displayed at the four test points, as shown above. The [Mag+Phs] flags show that,
• The two bearing blocks are moving in-phase with one another in the Y-direction (shaft direction)
• The two bearing blocks are moving 180 degrees out-of-phase with the motor in the Y-direction

WOBBLE OF THE INBOARD ROTOR

Because no unbalance screws were inserted in the two rotors when data for the ODS-FRFs in BLK: Balanced-1
were calculated, the machine “is assumed to be balanced”. However, the inboard rotor has a significant amount of
wobble in it.

Wobble of the inboard rotor creates an unbalance force in the Y-direction (shaft direction), resulting in a first-
order ODS with significant deflection in the Y-direction.

STEP 4 - COMPARING A BALANCED MACHINE ODS WITH A BASELINE ODS

In machinery health monitoring, (either route-based or in-situ), time waveforms and Auto spectra are typically
acquired and archived in a database. These archived measurements are then compared with prescribed warning
levels to ascertain the health of the machine. Vibration levels that exceed prescribed warning levels may require an
immediate or scheduled machine shut-down for maintenance.

When large amounts of vibration data are acquired from rotating machinery and archived in a database, the first
question should be asked, “How valid are those measurements?”

MAC & SDI will be used to compare different sets of ODS-FRFs with a Baseline set of ODS-FRFs.
• Press Hotkey 2 Import ODS-FRFs and select Balanced Case-1.BLK in the dialog box
• Press Hotkey 4 New Baseline ODS
When Hotkey 4 is pressed, the 12 M#s in BLK: Imported ODS-FRFs are copied into BLK: Baseline ODS-FRFs,
and the first-order ODS from BLK: Imported ODS-FRFs is displayed side-by side with the first-order ODS
from BLK: Baseline ODS-FRFs.
In the figure below, MAC & SDI ➔ 1 indicating that the first-order ODS in BLK: Imported ODS-FRFs is the
same as the first-order ODS in BLK: Baseline ODS-FRFs. This is expected since the M#s in BLK: Imported
ODS-FRFs replaced the M#s in BLK: Baseline ODS-FRFs.

Page 13 of 18
App Note 56 www.vibetech.com 6/1/2021

First-order ODS From Imported ODS-FRFs Compared to First-order ODS From BLK: Baseline ODS-FRFs.

TOOLS | M# PAIRS CORRELATION & MAGNITUDE RANKING

The display in the upper-right corner of the figure above is a Magnitude Ranking of SDI values that were stored
in a Shape Table when Tools | M# Pairs Correlation was executed between BLK: Imported ODS-FRFs and
BLK: Baseline ODS-FRFs. SDI was used to correlate the first-order ODS (Peak cursor values of the ODS DOFs)
in BLK: Imported ODS-FRFs with the first-order ODS in BLK: Baseline ODS-FRFs.

Since both Data Blocks contain the same M#s, all the bars in the Magnitude Ranking chart are SDI ➔ 1.0.

STEP 5 - COMPARE TWO DIFFERENT BALANCED CASE ODS’s

• Press Hotkey 2 Import ODS-FRFs and select Balanced Case-2.BLK in the dialog box
• Press Hotkey 4 Compare ODS with Baseline ODS
The figure below shows that BLK: Imported ODS-FRFs and BLK: Baseline ODS-FRFs contain similar first-
order ODS’s. The Magnitude Ranking also shows that all pairs of M#s with matching DOFs of the first-order
ODS’s in the two Data Blocks are similar.

Comparing the First-order ODS of Balanced Case-2 with Balanced Case-1.

Page 14 of 18
App Note 56 www.vibetech.com 6/1/2021

The Magnitude Ranking shows that all pairs of matching DOFs between the two Data Blocks have an SDI >
0.90.
MAC & SDI are evidence that Balanced Case-2.BLK & Balanced Cases-1.BLK contain repeatable
measurements from the balanced machine.

BALANCED CASE-3 VERSUS BALANCED CASE-1

• Press Hotkey 2 Import ODS-FRFs and select Balanced Case-3.BLK in the dialog box
• Press Hotkey 4 Compare ODS with Baseline ODS
Again, MAC & SDI in the figure below are evidence that Balanced-3.BLK & Balanced-1.BLK contain
repeatable measurements from the balanced machine.

Comparing the First-order ODS of Balanced Case-3 with Balanced Case-1.

All four Data Blocks for the balanced machine display a similar result to the one above,

STEP 6 - IMPORTING A BAD MEASUREMENT

To create a bad measurement, the tri-axial accelerometer was intentionally rotated 90 degrees on the outboard
bearing block, as shown below. Then data was acquired with the tri-axial accelerometer mounted incorrectly.
• BLK: Bad Measurement contains three bad ODS-FRFs for Point 1

Page 15 of 18
App Note 56 www.vibetech.com 6/1/2021

To compare the first-order ODS in BLK: Bad Measurement with the first-order ODS in BLK: Baseline ODS-
FRFs,
• Press Hotkey 2 Import ODS-FRFs and select Bad Measurement.BLK in the dialog box
• Press Hotkey 4 Compare ODS with Baseline ODS

MAC & SDI in the figure below indicate that the first-order ODS in BLK: Bad Measurement is different from
the first-order ODS in BLK: Baseline ODS-FRFs.
The Magnitude Ranking of SDI values in the upper right corner clearly indicates a significant difference between
the ODS’s for DOFs (1X, 1Y & 1Z).

Comparing the First-order ODS of a Bad Measurement at Point 1 with the BaseLine ODS.

MACHINE UNBALANCED CASES

Four unbalance cases were created by inserting screws in the rotors of the rotating machine. Then operating data was
acquired at four locations, and ODS-FRFs were calculated for each case.

Unbalance Created by Inserting Screws in the Machine Rotors.

Page 16 of 18
App Note 56 www.vibetech.com 6/1/2021

Unbalance was created by inserting a large screw into the inboard rotor and a small screw into the outboard rotor,
as shown below.

ODS-FRFs for each unbalance case are contained in the following Data Blocks
• Small screw on the outer rotor ➔ BLK: Unbalanced-Small Outer
• Large screw on the inner rotor ➔ BLK: Unbalanced-Large Inner
• Both screws 180 degrees apart ➔ BLK: Unbalanced-Both 180
• Both screws 0 degrees apart ➔ BLK: Unbalanced-Both 0

STEP 7 – IMPORTING THE WORST UNBALANCED CASE

• Press Hotkey 2 Import ODS-FRF and select Unbalanced-Both 0.BLK in the dialog box
Sine dwell animation of the first-order ODS will begin as shown below.
• Press Hotkey 3 ODS [Mag & Phs]

First-order ODS for Unbalanced Machine with Both Screws 0 Degrees Apart.

The [Mag & Phs] flags show that the motor is vibrating at over 0.77 IPS in the X-direction.

STEP 8 - COMPARING THE WORST UNBALANCE CASE ODS WITH A BASELINE ODS
• Press Hotkey 4 Compare ODS with Baseline ODS

In the figure below, MAC & SDI indicate that the first-order ODS of the unbalanced machine is quite different
from the first-order ODS of the balanced machine.

OVERLAYING THE ODS DISPLAYS

• Drag the scroll bar below the two models to the left
When the two models are overlaid, the difference between the two ODS’s is very apparent.

Page 17 of 18
App Note 56 www.vibetech.com 6/1/2021

First-order ODS’s of the Worst Case Unbalanced & Balanced Machine.

COMPARING ANY UNBALANCED CASE WITH ANY BALANCED CASE

The first-order ODS of any unbalance case can be compared with the first-order ODS of any balanced case
using the following Hotkeys
• Press Hotkey 2 and import a Data Block with ODS-FRFs for a balanced case
• Press Hotkey 4 to save the balanced case in BLK: Baseline ODS-FRFs
• Press Hotkey 2 and import a Data Block with ODS-FRFs for an unbalanced case
• Press Hotkey 5 to compare the first-order ODS for the unbalanced case with the balanced machine

REVIEW STEPS
To review the steps of this App Note,
• Press Hotkey 6 Review Steps

Page 18 of 18