MSC Nastran 2018.

MSC Nastran Rotordynamics User’s Guide

Rotordynamics User’s Guide

 Corporate Europe, Middle East, Africa
MSC Software Corporation MSC Software GmbH
4675 MacArthur Court, Suite 900 Am Moosfeld 13
Newport Beach, CA 92660 81829 Munich, Germany
Telephone: (714) 540-8900 Telephone: (49) 89 431 98 70
Toll Free Number: 1 855 672 7638 Email: [email protected]
Email: [email protected]

Japan Asia-Pacific
MSC Software Japan Ltd. MSC Software (S) Pte. Ltd.
Shinjuku First West 8F 100 Beach Road
23-7 Nishi Shinjuku #16-05 Shaw Tower
1-Chome, Shinjuku-Ku Singapore 189702
Tokyo 160-0023, JAPAN Telephone: 65-6272-0082
Telephone: (81) (3)-6911-1200 Email: [email protected]
Email: [email protected]

Worldwide Web
www.mscsoftware.com

Disclaimer
MSC Software Corporation reserves the right to make changes in specifications and other information contained
in this document without prior notice.
The concepts, methods, and examples presented in this text are for illustrative and educational purposes only,
and are not intended to be exhaustive or to apply to any particular engineering problem or design. MSC Software
Corporation assumes no liability or responsibility to any person or company for direct or indirect damages resulting
from the use of any information contained herein.
User Documentation: Copyright 2018 MSC Software Corporation. Printed in U.S.A. All Rights Reserved.
This notice shall be marked on any reproduction of this documentation, in whole or in part. Any reproduction or
distribution of this document, in whole or in part, without the prior written consent of MSC Software Corporation is
prohibited.
This software may contain certain third-party software that is protected by copyright and licensed from MSC
Software suppliers. Additional terms and conditions and/or notices may apply for certain third party software. Such
additional third party software terms and conditions and/or notices may be set forth in documentation and/or at
http://www.mscsoftware.com/thirdpartysoftware (or successor website designated by MSC from time to time).
PCGLSS 8.0, Copyright © 1992-2016, Computational Applications and System Integration Inc. All rights reserved.
PCGLSS 8.0 is licensed from Computational Applications and System Integration Inc.
MSC, Dytran, Marc, MSC Nastran, Patran, the MSC Software corporate logo, e-Xstream, Digimat, and Simulating
Reality are trademarks or registered trademarks of the MSC Software Corporation and/or its subsidiaries in the
United States and/or other countries.
NASTRAN is a registered trademark of NASA.FLEXlm and FlexNet Publisher are trademarks or registered
trademarks of Flexera Software. All other trademarks are the property of their respective owners.
Use, duplicate, or disclosure by the U.S. Government is subjected to restrictions as set forth in FAR 12.212
(Commercial Computer Software) and DFARS 227.7202 (Commercial Computer Software and Commercial
Computer Software Documentation), as applicable.
U.S. Patent 9,361,413
July 20, 2018
NA:V2018.2:Z:Z:Z:DC-ROTOR-PDF

Main Index
Documentation Feedback
At MSC Software, we strive to produce the highest quality documentation and welcome your feedback.
If you have comments or suggestions about our documentation, write to us at: documentation-
[email protected]
Please include the following information with your feedback:
 Document name
 Release/Version number
 Chapter/Section name
 Topic title (for Online Help)
 Brief description of the content (for example, incomplete/incorrect information, grammatical
errors, information that requires clarification or more details and so on.)
 Your suggestions for correcting/improving documentation
You may also provide your feedback about MSC Software documentation by taking a short 5-minute
survey at: http://msc-documentation.questionpro.com

Note: The above mentioned e-mail address is only for providing documentation specific
feedback. If you have any technical problems, issues, or queries, please contact Technical
Support.
 Contents
Rotordynamics User’s Guide
Contents

Contents

Preface
About this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Developments Introduced . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
List of MSC Nastran Books . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Technical Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Training and Internet Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1 Introduction
Introduction to Rotordynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Need for Rotordynamic analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Sources of Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Basic Concepts of Rotordynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Rotor Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Rotor Unbalance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Whirl of Simple Rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Asynchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Synchronous Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Campbell Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Mode Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Critical Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Torsional Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Overview of MSC Nastran Rotordynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2 Theoretical Formulation
Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Main Index
 2 Rotordynamics User’s Guide

Equations of Motion in Fixed Reference Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Damping in Stator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Damping in Rotor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Equation of Motion in Rotating Reference Frame. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Damping in Rotors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Damping in Stator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Final Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3 Elements Supported for Rotordynamics Analysis

Rotor Modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Rotordynamic Analysis using MSC Nastran. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Sample Test Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Model with ROTORG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Model using ROTORAX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Model with ROTOR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Rotor Modeling Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

4 Damping
Damping and Bearing Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Structural Damping Options for Stator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Damping Options for Rotating Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Circulation Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Effect of Circulation Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Controlling Rotor Damping and Circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Linear Bearings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Cross Coupling in Bearing Supports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Nonlinear Bearings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Frequency Dependent Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
THPAD using CBUSH2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Nonlinear Squeeze Film Damper, NLRSFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Equations for SFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Squeeze Film Damper in CBUSH2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Nonlinear Load Functions (NOLINi) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

5 Rotor Usage
Rotor Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Case Control Entries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
RGYRO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
CAMPBELL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Main Index
 Contents 3

Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
WR3, WR4, and WRH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
GYROAVG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
WHIRLOPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Bulk Data Entries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
RGYRO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
RSPINR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
RSPINT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
ROTHYBD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
CAMPBLL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
DDVAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
UNBALNC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

6 Rotordynamics with MSC Patran

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
MSC Patran Set up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Example Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Patran Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

7 Complex Eigenvalue Analysis for Rotors

Complex Eigenvalue Analysis for Rotors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Synchronous Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Asynchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Synchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Solvers available for complex eigenvalue analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Pardiso Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Mode Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Mode Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Mode Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Analysis in Fixed Reference Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Mode Tracking Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Analysis in Rotating Reference Frame. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Use of Whirl Parameter for Asynchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Analysis in Fixed and Rotating Reference Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Additional Options for Damping. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Effect of differential stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Summary for Equation of Motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

Main Index
 4 Rotordynamics User’s Guide

8 Frequency Response Analysis

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Asynchronous Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Synchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Equation used in the Analysis (Reference Frame) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Asynchronous Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Synchronous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Effect of Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Effect of GYROAVG parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Frequency Dependent Bearings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Modal Analysis (for ROTORG only) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161

9 Transient Response Analysis

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Equation of motion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Unbalance Loads for Transient Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Input file for transient analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Effect of External Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Effect of Internal Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Comparison with axisymmetric and 3D rotors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Usage of Nonlinear Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Mass and speed correction for Unbalance Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

10 Nonlinear Frequency Response

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
User Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
File Management Section (FMS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Executive Section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Case Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Bulk Data Entries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Analysis Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
Parameters for controlling the analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Handling Non-Convergence in Nonlinear Harmonic Response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Comments on Nonlinear Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Theoretical Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Example 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Example 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

Main Index
 Contents 5

11 Superelement Analysis with Rotors

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
PART Superelements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
External Superelements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Support for Copy and Move/Mirror Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Sample Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Input File for Creation Run . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Assembly Run. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Information in the Output File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
Results from External SE Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
SOL 107 Results (Complex Eigenvalue Analysis) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
SOL 108 Results (Direct Frequency Response) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Multilevel SE Setup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221

12 Multiple Rotors In Fixed System

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

A Axisymmetric Harmonic Elements for Rotordynamic Analysis

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Element Details. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Constraints/Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

B User Defined Subroutines Using CBUSH2D

Analysis in Windows Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Analysis on Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Compiling External CBUSH2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
Definitions Used in this document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Nomenclature
Glossary
References

Main Index
 6 Rotordynamics User’s Guide

Main Index
 Preface
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Preface


About this Book 2
 Organization 2

Developments Introduced 4

List of MSC Nastran Books 7
 Technical Support 8

Training and Internet Resources 8

Main Index
 2 MSC Nastran Rotordynamics User’s Guide
About this Book

About this Book

MSC Nastran is a general purpose finite element program which solves a wide variety of engineering
problems. It is developed, marketed, and supported by the MSC Software Corporation.
The MSC Nastran Rotordynamics User's Guide is intended to explain the capability of MSC Nastran to
perform rotordynamic analysis as available in MSC Nastran V2018. Prior to this user`s guide, the overview
of updates made in rotordynamic analysis were provided through the Release Guide associated with each
version. The rotordynamic analysis is used to obtain complex eigenvalues, determine stability and to perform
frequency response and transient response analysis for a system where one or more components are rotating.
Guidance is provided in preparing the input file and assessing the output. A number of examples are provided
to illustrate the use of MSC Nastran`s Rotordynamics features.

Organization
The guide is divided into twelve chapters and three Appendices. A brief description of each of these is:

Chapter 1. Introduction to Rotordynamics

This chapter provides an overview of rotordynamic analysis and briefly explains important concepts. For a
more detailed description, readers are referred to books and papers listed in Appendix A of this manual. This
chapter assumes that the reader has basic knowledge of the field and general concepts related to structural
dynamics.

Chapter 2. Equations of Motion

This chapter provides theoretical overview of rotordynamics. Using a simplified Jeffcott rotor model,
equations of motion are obtained for the model in fixed and rotating reference frames using Lagrange’s
principle. The effect of damping is also included in the analysis. The equations of motion generated in this
chapter form the backbone for the analysis presented in rest of the book.

Chapter 3. Rotor Modeling

Different options available in MSC Nastran for modeling a rotor are described in this chapter. Based on the
required fidelity and complexity of rotor model, the user has the freedom to choose the appropriate modeling
technique. This chapter also identifies connectors that can be used for connecting a rotor to another rotor or
to a stator.

Chapter 4. Damping and Rotor Bearing Models

MSC Nastran supports various options for specifying damping in dynamic analysis. This chapter highlights
usage of these damping options for rotors and stators and shows the corresponding Bulk Data entry. Next,
different nonlinear elastic elements, which can be used to model the complex behavior of bearings used in
rotordynamic analysis, are described.

Main Index
 Preface 3
Organization

Chapter 5: Rotor Usage

This chapters lists various rotordynamic related entries available in MSC Nastran to define the components
of a rotor and enable different kinds of analyses. These entries may be defined in File Management Section,
Case Control Section or in the Bulk Data Section of the input file.

Chapter 6: Rotordynamics using Patran

Currently, MSC Patran provides limited support for rotordynamics. This chapter provides step-by-step
instruction for creating the first rotordynamic analyses model using Patran. Later, post-processing options
available to plot the modeshapes and to generate a Campbell diagram are discussed.

Chapter 7: Complex Eigenvalue Analysis

This chapter describes the usage of complex eigenvalue analysis to obtain whirl speeds and critical speeds
using analysis in asynchronous and synchronous mode, respectively. The analyses is performed in both fixed
and rotating coordinate system.

Chapter 8: Frequency response analysis

In rotordynamics analysis, asynchronous or synchronous forcing function can be defined in order to study
their effects on rotors. MSC Nastran includes separate bulk data entry to define the most commonly used
synchronous load due to unbalance. This chapter also highlights the effect of damping and circulation terms
on rotor response.

Chapter 9: Transient response Analysis

The transient analyses in MSC Nastran allows the usage of nonlinear bearings and it includes effect of rate
of change of rotor speed.

Chapter 10: Nonlinear frequency response analysis

The SOL 128 in MSC Nastran allows for mild nonlinearity to be included in the frequency response analysis.
In this case, the problem is solved using the harmonic balance technique.

Chapter 11: External SE analysis with rotors

This chapter describes the steps involved in defining multiple rotors in multiple external superelements to
streamline the analysis for very large models. Besides these, identical components of a model which are
repeated multiple times, can be generated using copy/move or mirror/move approach to reduce the
preprocessing and analyses time.

Chapter 12: Multiple Rotors

MSC Nastran allows for multiple rotors to be defined in the analysis in fixed reference frame where each rotor
is spinning at a different speed. Using an example, this chapter describes the steps involved in defining the
model and interpreting the results.

Main Index
 4 MSC Nastran Rotordynamics User’s Guide
Developments Introduced

Appendix A. Element Details for Axisymmetric Harmonic Elements

This appendix provides details related to element formulation for axisymmetric harmonic elements for
rotordynamic analysis.

Appendix B. User Defined Subroutines Using CBUSH2D

This appendix describes steps involved in defining a user defined subroutine for usage in CBUSH2D entry.

Developments Introduced
New versions of MSC Nastran are released periodically. Details about the updates made in each released
version are described in the Release Guide provided with MSC Nastran. Rotordynamics capability was first
introduced in the MSC Nastran 2004 release. Since then significant progress have been made, and new
features have been added with each release based on customer requirements and key features identified. The
major enhancements since the MSC Nastran 2004 version are listed here:

MSC Nastran 2004

 Introduced rotordynamic analysis in MSC Nastran for 1-D models defined using
CONM2/CBAR/CBEAM elements. Components of this 1-D rotor are defined using a new Bulk
Data entry called ROTORG and rotor usage is defined using RGYRO and RSPINR.

MSC Nastran 2005 r3

 Introduced squeeze film damper. The squeeze film damper model is based on work originally
performed at Case Western Reserve University (CWRU).

2006 MD Nastran
 Additional damping options for rotors (structural, Rayleigh, and hybrid damping) in RSPINR and
RSPINT.
 NLRSFD implemented as an element.
 The rotordynamics option is available in the Aerodynamic Flutter (SOL 145) and Aeroelastic Gust
 Response (SOL 146) solutions. SOL 145 uses the same gyroscopic equations as Complex
Eigenvalues; SOL 146 uses the frequency response formulation.
 Generation of Campbell diagrams using CAMPBLL entry.
 Preliminary rotor mode tracking.
 Updated equations of motion for complex eigenvalues analysis, frequency response, and transient
response by including damping and circulation terms for rotors.

2007 MD Nastran R2.1

 Introduced unbalance loading for frequency response.
 Nonlinear transient rotordynamics is introduced in SOL 400.

Main Index
 Preface 5
Developments Introduced

2008 MD Nastran R3
 In this version, special options were added to SOL 106 to support blade vibration analysis.
 Integrated, Multi-disciplinary Fan Blade Out (FBO) and Rotordynamics simulation in SOL 700.
 Allow for user defined subroutines so the user can provide their own subroutines or applications
within an MSC Nastran execution process. For example, the user can provide their own formulation
of a squeeze film damper in rotordynamic analysis.

2010 MD Nastran
 Nonlinear harmonic response for rotordynamic analysis is introduced. It is helpful for analyzing a
class of dynamic response problems where rotor/stator contact under abnormal running conditions.

2011 MD Nastran
 Starting with MD Nastran 2011, the user is allowed to provide their own SCA-object to calculate
the properties of CBUSH2D elements.
 Introduced ROMAC’s (University of Virginia Rotating Machinery And Controls laboratory,
http://www.virginia.edu/romac/) THPAD routine (which must be obtained from ROMAC) to
calculate the properties of a tilting pad journal bearing.

2012 MSC Nastran

 Introduced IRAM as a new complex eigenvalue extraction method in SOL 107 analysis for
rotordynamic analysis.

2013 MSC Nastran

 Allows usage of axisymmetric harmonic elements for rotordynamic analysis using a new bulk data
entry called “ROTORAX”. This allows for a better capture of the geometry of the rotors, and thus
leads to better and more accurate predictions of the behavior of the rotordynamic models.
 A new entry called “RBAX3D” is introduced to connect an axisymmetric harmonic grid point to a
3-D grid point.
 A limited capability to allow rotors in external superelements is provided.

2014 MSC Nastran

 In the 2014 release, we augmented the previous capability to allow concentrated masses to be used in
an axisymmetric harmonic calculation.
 With the enhancements in MSC Nastran 2014, any number of ROTORG, ROTORSE, and/or
ROTORAX Bulk Data entries that define rotors may be specified in external SEs and the residual.

2014.1 MSC Nastran

 Added an option to apply Rayleigh damping for rotors as it was implemented in MSC Nastran
2005.

Main Index
 6 MSC Nastran Rotordynamics User’s Guide
Developments Introduced

2016 MSC Nastran release

 A new bulk data entry “ROTOR” is introduced to allow for 3-D elements (solid and shells) for
analysis in fixed and rotating reference frames.
 The user can now optionally suppress the effect of the circulation matrix for damping defined in
rotors for analysis in a fixed reference frame.
 For 3-D rotors in a rotating reference frame, effects of stress stiffening (differential stiffness) can be
incorporated in the analysis.

2017 MSC Nastran

 Introduced option to define WR3, WR4 and WRH parameters through RSPINR/RSPINT entries
to define these parameters individually for each rotor.
 Added CFLAG option in UNBALNC entry to include mass and speed correction in transient
analysis in fixed reference frame.
 Campbell Diagram option is now supported in SOL 110 for all the rotors. In addition, modal
solution sequences, SOL 110 and SOL 111, can now be used for analysis of 2D and 3D rotors.

2018 MSC Nastran

 Introduced two mode tracking algorithms for Campbell diagram:
• Based on first and second derivative continuity for variation of frequency with rotor speed.
• Based on orthogonality condition for complex left and right eigenvectors.
 Improved output from complex eigenvalue analysis by identifying rotor whirl type for each rotor in
the model for both SYNC and ASYNC analysis.
 Enhanced PBUSH2D entry by introducing option to specify cross coupling terms.
 Allow for one or more rotors in multilevel Part SE.
 Fix rotor spin direction issue.

Main Index
 Preface 7
List of MSC Nastran Books

List of MSC Nastran Books

A list of some of the MSC Nastran documents is as follows:

Installation and Release Guides

 Installation and Operations Guide
 Release Guide
Guides
Reference Books
 Quick Reference Guide
 DMAP Programmer’s Guide
 Reference Guide
 Utilities Guide
User’s Guides
 Getting Started
 Linear Static Analysis
 Dynamic Analysis
 Embedded Fatigue
 Embedded Vibration Fatigue
 Demonstration Problems
 Thermal Analysis
 Superelements
 Design Sensitivity and Optimization
 Rotordynamics
 Implicit Nonlinear (SOL 600)
 Explicit Nonlinear (SOL 700)
 Aeroelastic Analysis
 User Defined Services
 Non Linear (SOL 400)
 High Performance Computing
 DEMATD

You may find any of these documents from MSC Software at:
http://simcompanion.mscsoftware.com/infocenter/index?page=home

Main Index
 8 MSC Nastran Rotordynamics User’s Guide
Technical Support

Technical Support
For technical support phone numbers and contact information, please visit:
http://www.mscsoftware.com/Contents/Services/Technical-Support/Contact-Technical-Support.aspx
Support Center (http://simcompanion.mscsoftware.com)
The SimCompanion link above gives you access to the wealth of resources for MSC Software products. Here
you will find product and support contact information, product documentations, knowledge base articles,
product error list, knowledge base articles, and SimAcademy Webinars. It is a searchable database which
allows you to find articles relevant to your inquiry. Valid MSC customer entitlement and login is required to
access the database and documents. It is a single sign-on that gives you access to product documentation for
complete list of products from MSC Software, allows you to manage your support cases, and participate in
our discussion forums.

Training and Internet Resources

MSC Software (www.mscsoftware.com)
MSC Software corporate site with information on the latest events, products and services for the
CAD/CAE/CAM marketplace.

http://simcompanion.mscsoftware.com
The SimCompanion link above gives you access to the wealth of resources for MSC Software products. Here
you will find product and support contact information, product documentations, knowledge base articles,
product error list, knowledge base articles, and SimAcademy Webinars. It is a searchable database which
allows you to find articles relevant to your inquiry. Valid MSC customer entitlement and login is required to
access the database and documents. It is a single sign-on that gives you access to product documentation for
complete list of products from MSC Software, allows you to manage your support cases, and participate in
our discussion forums.

http://www.mscsoftware.com/msc-training
The MSC-Training link above points you to schedule and description of MSC Seminars. The following
courses are recommended for beginning MSC Nastran users:

NAS101A - Linear Static and Normal Modes Analysis using MSC Nastran
This course serves as an introduction to finite element analysis. It includes discussion of basic features
available in MSC Nastran for solving structural engineering problems. In this course, all finite element
models will be created and edited using a text editor, not a graphical pre-processor. Proper data structure of
the MSC Nastran input file is covered. At the conclusion of seminar, the student will be familiar with
fundamental usage of MSC Nastran.

Main Index
 Preface 9
Training and Internet Resources

NAS101B - Advanced Linear Analysis using MSC Nastran

This course is a continuation of NAS101A - Linear Static and Normal Modes Analysis using MSC Nastran.
In this class, you will learn:
• Theory of buckling analysis and how to perform a buckling analysis
• About rigid elements - MPC, RBAR,RBE2, and RBE3
• Modeling with interface element CINTC and connectors
• Lamination theory and composite materials
• MSC Nastran composite theory
• Failure theories
• Linear contact and permanent glued contact
• Different model checks
• Modeling tips and tricks

NAS120 - Linear Static Analysis using MSC Nastran and Patran

This seminar introduces basic finite element analysis techniques for linear static, normal modes, and buckling
analysis of structures using MSC Nastran and Patran. MSC Nastran data structure, the element library,
modeling practices, model validation, and guidelines for efficient solutions are discussed and illustrated with
examples and workshops. Patran is an integral part of the examples and workshops and is used to generate
and verify illustrative MSC Nastran models, manage analysis submission requests, and visualize results. This
seminar provides the foundation required for intermediate and advanced MSC Nastran applications.

NAS122 - Basic Dynamic Analysis using MSC Nastran and Patran

The course covers a wide range of dynamic analysis topics from basic to advanced using an integrated
approach. Patran is used for data set up and postprocessing and MSC Nastran is used for the solver. Many
unique practical hints and tips are given which do not exist in other material. Case studies are used in each
topic to help understand the physics and engineering behind the techniques in a practical way. A
comprehensive set of over 20 fully detailed student workshops is used to obtain real "hands on" experience.
A strong emphasis is placed on engineering process so that the student can rapidly relate the course to his or
her project needs.

NAS127 - Rotordynamic Analysis using MSC Nastran

This course covers rotordynamic analysis for coupled rotating and stationary components like jet engines,
turbines, compressors, energy storage devices, etc. This course provides details for:
 How to setup and analyze structural models with one or more rotating components
 The types of analysis supported by the rotordynamics capability
 Static analysis
 Complex eigenvalue analysis (modal and direct)
 Frequency response (modal and direct) and nonlinear frequency response
 Transient response (direct linear and nonlinear)

Main Index
 10 MSC Nastran Rotordynamics User’s Guide
Training and Internet Resources

 Damping effects and input methods for models with rotating components
 How to use Patran to create models and display results with animation or graphs
There are two sets of workshops for this course. One set uses Patran as the pre and postprocessor. The other
directly edits the MSC Nastran input file.

Main Index
 Chapter 1: Introduction
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

1 Introduction


Introduction to Rotordynamics 12

Basic Concepts of Rotordynamics 13
 Overview of MSC Nastran Rotordynamics 22

Main Index
 12 MSC Nastran Rotordynamics User’s Guide
Introduction to Rotordynamics

Introduction to Rotordynamics
Rotordynamics is that branch of system dynamics dealing with mechanical devices in which at least one part,
usually defined as a rotor, rotates with significant angular momentum (Genta, 2007). Rotating systems have
been around for many centuries, and a broad historical perspective of this subject is provided in a book by
Rao (Rao, 2011)
Rotating machines represent large class of machinery used in industrial applications; for example, power
stations, marine propulsion systems, aircraft engines, machine tools, transportation systems, and various
other purposes. Rotors equipped with bladed disks or impellers, rotating at high speeds in a fluid
environment allow rotating machines to produce, absorb, transform, or condition an amazing amount of
energy often in comparatively small, compact packages. The safe and sound operation of such rotating
machinery is of paramount importance. Every rotating machine consists of three principal components: the
rotor, the bearings, and their support structure. The rotor is connected to the support structure by means of
bearings, which may be of several types.
Of all the dynamic problems associated with rotating machinery, excessive vibration and instability are very
important, and the endeavor of the designer is to restrict the vibration of such systems as much as possible
and to increase the zone of stable operation. These problems unless avoided, lead to excessive vibration, wear,
operational difficulties, or even catastrophic failures. All three important modes of vibrations (lateral,
torsional, and axial modes) may be present during rotor operation. Among these modes, the lateral modes of
the rotor are of great concern, as they represent the lowest modes in most of the cases.

Need for Rotordynamic analysis

Rotordynamics is different from structural vibrations analysis because of gyroscopic moments, cross-coupled
forces, and the possibility of whirling instability (Vance, 2010). The power density in a rotating machine
strongly affects its rotordynamics. In the 1920s, the turbine industry designed machines to operate at
substantially higher loads and at speeds above the lowest critical speed, and this introduced modern
Rotordynamics problems.
In designing, operating, and troubleshooting rotating machinery, rotordynamics analysis can help accomplish
the following objectives:
 Predict critical speeds.
 Determine design modifications to change critical speeds.
 Predict natural frequencies of torsional vibration.
 Calculate balance correction masses and locations from measured vibration data. (not currently
supported in MSC Nastran)
 Predict amplitudes of synchronous vibration caused by rotor imbalance. (using frequency response
or transient analysis)
 Predict frequency or transient response to externally applied synchronous/asynchronous force.
 Predict threshold speeds and vibration frequencies for dynamic instability.
 Determine design modifications to suppress dynamic instabilities.
 Study effect of external features like rub/friction/aerodynamics.

Main Index
 Chapter 1: Introduction 13
Basic Concepts of Rotordynamics

Sources of Vibration
A rotor can vibrate in three ways: axially, torsionally and laterally. Lateral vibrations cause displacements of
the rotor: normal to the axis of rotation. These lateral motions combine to produce an orbital motion of the
rotor in a plane perpendicular to its axis of rotation. Rotor lateral vibration is usually the largest component
of vibration in most high-speed machinery. Excessive lateral vibration can cause damage to both bearings and
support structure.
Rotor unbalance is the first and best known contributing factor to vibration of a rotor. When the rotor mass
center line does not coincide with its rotational axis, then mass unbalanced inertia related rotating forces
occur. As this rotates together with the rotor and is oriented perpendicular to the rotational axis, the
unbalance excites the lateral vibration modes. Higher speeds cause much greater centripetal forces due to
imbalances, which are a function of the square of the operating frequency. The forces are usually periodic,
and therefore, produce periodic displacements observed as vibration (Vance, 2010).
The frequency of the vibration response corresponds to the frequency of the excitation force. The frequency
of unbalance response is the same as the rotational speed; thus, the unbalance related synchronous lateral
vibrations are referred to as 1 X vibrations. When the rotor system is nonlinear, more frequency components
may be excited. The corresponding frequencies usually represent multiples of the excitation frequency,
superharmonics (2 X, 3 X, and so on) and subharmonics(1/2 X, 1/3 X and so on). Another class of excitations
are “free vibrations” or “transient vibrations”, which occur when the system is excited by a short duration
impact. There also exists a third category, self-excited vibrations. These vibrations are sustained by a constant
source of energy, which may be external or internal.

Basic Concepts of Rotordynamics

Figure 1-1 Simplified Jeffcott Rotor

Main Index
 14 MSC Nastran Rotordynamics User’s Guide
Basic Concepts of Rotordynamics

Many of the concepts of rotordynamics can be explained using a simplified Jeffcott rotor (Jeffcott, 1919) as
shown in Figure 1-1. Consider a massless bar attached to springs and dampers at its ends A and B. The bar is
rotating about the Z axis oriented along its length. The X and Y axes represent lateral bending directions. A
disk with finite mass and inertia is attached to the bar at the center. The rotating bar is supported at the ends
by bearings which are modeled using linear springs and dampers. Various terms used in Figure 1-1 are
described here:
KXR and KYR Bending stiffness of the bar in the X and Y directions, respectively.
KXA and KYA Bending stiffness of the bearing at end A in the X and Y directions, respectively.
KXB and KYB Bending stiffness of the bearing at end B in the X and Y directions, respectively.
CXA and CYA Damping coefficients of the bearing at end A in the X and Y directions, respectively.
CXB and CYB Damping coefficients of the bearing at end B in the X and Y directions, respectively.
m Mass of the disk.
IP, IdX, IdY Polar and diametral moment of inertia for the disk.

Reference frame
A rotordynamic problem can be analyzed either in a fixed (inertial) reference frame or in a rotating coordinate
reference frame. Depending upon the symmetry present in the rotor and stator, an appropriate reference
frame can be selected for analysis to obtain time-independent equations of motion. When the rotor, bearings,
and support structure contain asymmetry, then the resulting equations contain time-dependent parameters
independent of the reference frame employed for analysis.

Figure 1-2 Fixed and Rotating Coordinate System

The general transformation of a time-dependent vector from a stationary coordinate system to a rotating
coordinate system at the same origin rotating at Ω rad/sec may be written as

{ v p ( t ) } = [ T ( Ω, t ) ] { v p ( t ) }

Where:

Main Index
 Chapter 1: Introduction 15
Basic Concepts of Rotordynamics

{ vp ( t ) } = rotating coordinate system definition of time-dependent vector.

[ T ( Ω, t ) ] = time-dependent transformation matrix from rotating to stationary system.
{ vp ( t ) } = stationary coordinate system definition of time-dependent vector.

Transformation from fixed to rotating coordinate system is given by:

cos Ωt sin Ωt 0
{ vp ( t ) } = – sin Ωt cos Ωt 0 { v p ( t ) }
0 0 1

The key differences between the analysis in a fixed system and the analysis in a rotating system are given
below:

Table 1-1 Difference between Analysis in Fixed and Rotating System

Fixed System Analysis Rotating System Analysis
Rotating part of the structure must be axisymmetric Rotating structure need not be axisymmetric
about the spin axis. about the spin axis.
Rotating structure can be connected to a stationary Only symmetric non-rotating structure is
structure in the analysis model. included in the analysis.
Supports more than one rotating structure spinning at Supports only a single rotating structure.
different rotation speeds about different axes of
rotation.
Can be used to generate Campbell diagrams for Campbell diagrams are not directly applicable for
computing rotor critical speeds. computing critical speeds.

Rotor Symmetry
In order to determine whether to perform the analysis in fixed or rotating reference frames, it is important to
understand if the rotor and support structure are symmetric or not. For the simplified Jeffcott rotor shown
in Figure 1-1
The rotor is considered to be symmetric if all of the following conditions are met:
KXR = KYR
IdX = IdY
And the stator is considered to be symmetric if all of the following conditions are met:
KXA = KYA
CXA = CYA
KXB = KYB
CXB = CYB

Main Index
 16 MSC Nastran Rotordynamics User’s Guide
Basic Concepts of Rotordynamics

In order to work with time-independent mass, stiffness and damping matrices in the equations of motion,
the following conditions must be satisfied.

Table 1-2 Analysis Reference Frame

Reference Case Rotor Symmetry Stator Symmetric Analysis Frame
Case 1 Symmetric Symmetric Fixed/Rotating
Case 2 Symmetric Not Symmetric Fixed Only
Case 3 Not Symmetric Symmetric Rotating Only
Case 4 Not Symmetric Not Symmetric NA

Table 1-2 provides summary with regard to reference frame that can be used for the analysis, depending on the
symmetry in rotor and stator, in order to obtain time-independent system of equations. In Case 1, both the
rotor and stator portion of the model are symmetric and the analysis can be performed in any reference frame.
If the rotor is symmetric and the supporting structure is asymmetric, the analysis can be performed in fixed
reference frame only (most common scenario). In case the rotor is asymmetric and the stator is symmetric,
the analysis can be performed in rotating reference frame only. And finally, if both the rotor and stator are
asymmetric, it is not possible to obtained time-independent system of equations. This analysis is currently
not supported in MSC Nastran.

Rotor Unbalance
As mentioned earlier, rotor unbalance is the first and best known method contributing to vibration of a rotor.
When the rotor mass center line does not coincide with its rotational axis, then mass unbalanced inertia
related rotating forces occur. As this rotates together with the rotor and is oriented perpendicular to the
rotational axis, the unbalance acts in lateral vibration mode. Higher speeds cause much greater centripetal
forces due to imbalances, which is a function of the square of the operating frequency. The frequency of the
vibration response corresponds to the frequency of the excitation force. The frequency of unbalance response
will be the same as the rotational speed; thus the unbalance related synchronous lateral vibrations are referred
to as 1 X vibrations. Residual unbalance can occur due to various causes such as manufacturing error, thermal
deformation, material inhomogeneity, wear, and corrosion. Unbalance can also occur due to a stack up of
tolerance in assembly.

Whirl of Simple Rotor

A planar motion of a rotor is called a whirling motion or a whirl. Understanding the whirling phenomena of
rotors is essential for a deep insight into the physics of rotating machinery. The simple model considered here
is directly applicable to practical rotor problems observed in the real world and the equations of motion
governing the simple model can be easily understood.
The analysis presented here uses the Jeffcott rotor shown in Figure 1-1, the simplest rotor model. The Jeffcott
rotor model is allowed to vibrate simultaneously in both directions (vertical and horizontal) producing a whirl

Main Index
 Chapter 1: Introduction 17
Basic Concepts of Rotordynamics

orbit. The Jeffcott rotor undergoing whirl motion is shown in Figure 1-3. The front view of the rotor model
is also shown where a reference point P on the disk circumference is highlighted. The rotor disk assembly is
spinning in the anti-clockwise direction with a fixed angular velocity. Depending upon the applied force, two
different whirl modes are possible: forward and backward whirl. As shown in Figure 1-4, in the case of forward
whirl motion, the rotor assembly whirls in the same direction as the applied angular velocity. Whereas in the
case of backward whirl motion, the rotor whirl direction is opposite to that of the applied angular velocity.

Figure 1-3 Whirl orbit for rotor

Figure 1-4 Forward and backward whirl direction

Stability
Whirl frequencies are important in determining rotor stability. Although the whirling motion may not be at
the rotational speed, whirl modes may be unstable due to internal rotor damping or external forces due to
bearing and seals. Estimating the stability of the whirl modes is important to ensure that a system operates
smoothly.
The eigenvalues of the system appear in complex conjugate pairs in the form

Main Index
 18 MSC Nastran Rotordynamics User’s Guide
Basic Concepts of Rotordynamics

p = α ± iω

where α is the damping exponent (rate of growth or decay of the modal response) and ω is the damped natural
frequency. An eigenvector, or a mode shape, corresponding to every eigenvalue can be computed. The mode
shapes also appear in complex conjugate pairs. Each complex mode represents a time-dependent curve in
space representing the rotor deformation at the corresponding whirling frequency.
System stability is investigated by checking that all eigenvalues have negative real parts; i.e., all roots of the
system characteristic equation lie in the left-hand side of the p-plane (plot with α on the x-axis and ω on the
y-axis). By varying some parameter in the rotor model, such as the rotor spin speed or the damping coefficient
at one of the bearings and computing the eigenvalues at the different parameter values, stability analysis can
be performed with respect to that parameter using a root locus plot.

Asynchronous Analysis
Calculation of response with the rotors spinning at set speeds is performed using the asynchronous option.
The user specifies the rotor speeds, the gyroscopic terms are determined and added to the equation of motion,
and the response is determined.

Synchronous Analysis
Calculation of frequency response with the excitation synchronous with the reference rotor is performed
using the synchronous option. The user specifies the rotor speed range, the gyroscopic terms are determined
for each frequency and added to the equation of motion, and the response is determined. If complex
eigenvalue analysis is performed with the “synchronous” option, the critical frequencies of the model are
obtained as output.

Campbell Diagram
For whirl frequencies analysis, the rotational speeds are specified as input, and the whirl frequencies are
computed at those speeds. Typically, a series of analyses at different rotor speeds is performed in complex or
modal complex eigenvalue solutions. It is common practice to plot the whirl frequencies versus the rotational
speed of the rotor as shown in Figure 1-5. This plot is called a Campbell diagram. Whirl frequencies for each
mode at different rotor speeds are connected to form a series of lines that represent the natural frequency
curves.

Main Index
 Chapter 1: Introduction 19
Basic Concepts of Rotordynamics

Figure 1-5 Campbell diagram for a Rotor

Mode Tracking
For Campbell diagram plotting, it is often observed that two modes may switch order depending upon their
dependence on rotor speed, as shown in Figure 1-6(a). This may lead to confusion while analyzing modes at
higher rotor speeds. In order to avoid this, a mode tracking algorithm is required which tracks the mode shape
for each eigenfrequency. Results obtained after using a mode tracking algorithm are shown in Figure 1-6(b).
Most of the mode tracking algorithms make use of orthogonality conditions. A robust approach for mode
tracking is currently under development and will be included in future version of MSC Nastran.

Main Index
 20 MSC Nastran Rotordynamics User’s Guide
Basic Concepts of Rotordynamics

Figure 1-6 Mode Tracking for Campbell Diagram

Critical Speed
Critical speeds are commonly defined as the rotational speeds at which vibration due to rotor unbalance is a
local maximum. Since rotor unbalance always produces a synchronous whirling, large amplitudes of
synchronous vibration usually indicate a rotor unbalance problem. However, not all unbalance responses at
critical speeds are dangerous. There can be other rotational speeds at which subsynchronous or
supersynchronous excitation sources can produce large destructive vibration; these are called subsynchronous
and supersynchronous critical speeds.
Campbell diagrams can be used to determine the rotor critical speeds, as shown in Figure 1-7. A critical speed
of a rotor system is defined as a rotational speed of the system when a multiple of that speed coincides with
one of the system's natural frequencies. Excitation due to unbalance is usually synchronous; i.e. the rotational
speed of the rotor is equal to one of the natural frequencies. Excitation of superharmonics or subharmonics
occurs at rotational speeds that are multiples or fractions of these frequencies, respectively. One approach for
determining critical speeds is to simply generate the Campbell diagram, include all excitation frequency lines
of interest, and graphically find the intersection between the excitation line and the natural frequency curves
to define the critical speed associated with that excitation.

Main Index
 Chapter 1: Introduction 21
Basic Concepts of Rotordynamics

Figure 1-7 Critical Speed for a Rotor

Torsional Vibration
Individual turbomachinery rotors are generally stiff enough in torsion to put their natural frequencies of
torsional vibration above the range of most torsional excitations. (Vance, 2010)
Key differences between lateral rotor vibration and torsional vibration are listed here:
 Lateral vibration is easily detected through standard instrumentation or through vibrations
transmitted to housings and foundations. Also, large amplitudes of lateral vibration are often noticed
due to rubbing of rotating seals and process wheels. On the other hand, instrumentation for
torsional vibration is not usually installed, and large amplitudes can occur silently and without much
effect on housings and foundations.
 Natural frequencies of lateral vibration are influenced by rotating speed, whereas natural frequencies
of torsional vibration are independent of rotating speed and can be measured with the machine at
rest if excitation can be provided.
 Lateral vibration in rotating machines can become unstable; this is very rare for torsional vibration in
machines without speed control feedback.
 The most common excitation of lateral vibration is synchronous (1 X) from rotor imbalance. Rotor
imbalance has no effect on torsional vibration, except indirectly in machines with gears where the
lateral vibration produces dynamic torque.

Main Index
 22 MSC Nastran Rotordynamics User’s Guide
Overview of MSC Nastran Rotordynamics

Overview of MSC Nastran Rotordynamics

The rotordynamics capability provides users with a relatively simple method of performing the design and
analysis of structures with rotating components. The procedures were developed with input from aircraft
engine manufacturers to ensure that their basic requirements were met. The rotordynamic capability in MSC
Nastran can be used in frequency response (direct and modal), complex modes (direct and modal), static,
nonlinear transient, and linear transient (direct only) analyses.
Frequency response analyses determine the rotor/support response to arbitrary excitation with the rotors
spinning at user specified rates or excitation that is synchronous with the reference rotor. The asynchronous
calculation determines the response of the system to an external input that is independent of the rotor speed.
The synchronous calculation determines the system response to a rotor imbalance or other excitation that is
dependent on the rotor spin rate.
Complex modes analyses determine the whirl frequencies and critical speeds. Whirl modes are modes of a
rotor/support system with the rotors spinning at specified rates. Critical speeds are whirl frequencies that
coincide with a rotor spin rate. Calculation of whirl frequencies is performed by the asynchronous option
(ASYNC). Calculation of critical speeds is performed by the synchronous option (SYNC).
Static analysis determines, for example, the loads on the rotor due to pitch or yaw of an aircraft. Severe
transverse rotations may be part of the design environment and limiting rotor motion to prevent rotor blades
from rubbing against the interior of the casing may be design criteria.
Direct linear and nonlinear transient analyses allow simulation of engine blade-out and subsequent
windmilling. The analysis is used to ensure structural integrity during flight and guarantee that excessive
vibration levels are not reached.
In rotordynamics analyses, very often bearing and dampers exhibit nonlinear behavior. If these were to be
analyzed accurately using transient analysis, it may require significant computational time. For these kind of
problems, where the model shows mild nonlinearity, nonlinear frequency response analysis can be performed
to approximately account for the presence of nonlinearity. This kind of approach is very helpful in the initial
design cycle.
When the problem is being analyzed in the fixed reference frame, the rotating part of the model needs to be
symmetric about the axis of rotation. In case the rotor being analyzed is NOT symmetric, then the problem
needs to be solved in the rotating reference frame. However, in this case, the nonrotating portion needs to be
symmetric in order to obtain equations of motion with constant coefficients.
With the enhanced 3D Rotordynamics capability, although it might take more computational time with 3D
shell/solid elements to model the 3D rotor(s), it is much easy to capture the exact geometry without special
modeling constraints or efforts, which would greatly reduce the time for model preparation. The 3D
modelling capability also provides more information about the model by capturing the local deformation
modes. For simpler, symmetric and uniform models usage of 1D beam/bar elements is encouraged since they
provide significant advantage in computational time. The use of axisymmetric harmonic elements is
recommended when the rotor is symmetric and it includes tapered/Nonuniform sections.
All the above analyses can be performed in the stationary (inertial) coordinate system with the assumption
that the rotors have rotational symmetry or in a rotating coordinate system with the assumption that the
stator has rotational symmetry.

Main Index
 Chapter 1: Introduction 23
Overview of MSC Nastran Rotordynamics

Benefits
Traditionally, companies have used proprietary in-house codes and customized DMAP solutions to perform
analyzes of structures with rotating components. The rotordynamics capability in MSC Nastran provides an
easy-to-use standardized procedure for analyzing rotating structures. The rotordynamics capability is
included in the standard MSC Nastran solution sequences, and invoked using case control commands and
bulk data entries. This standardization allows easier exchange of model data between in-house groups or
external vendors.
The design of the rotordynamics capability allows multiple rotors to be included in the analyses. The rotors
may spin at different rates and be aligned at arbitrary angles.
MSC Nastran rotordynamics supports analysis using 1-D elements, axisymmetric harmonic elements and 3-
D solid and shell elements in fixed and rotating reference frames. Based on the complexity of the model and
required fidelity, the user can select the appropriate modeling techniques.
The enhanced external superelement (SE) capability introduced in MSC Nastran 2014 allows the user to
have multiple rotors in multiple external SEs as well as in the residual and to employ them to form complete
rotordynamic configurations in subsequent assembly jobs for performing specific types of analysis similar to
complex eigenvalue analysis or frequency or transient response analysis. This feature also provides the
capability for handling and managing external SEs that are geometrically identical by copying and
moving/mirroring a primary external SE to generate secondary external SEs. This can greatly enhance user
convenience and productivity by eliminating the need to model geometrically identical external SEs
individually.
There are benefits of different types of modeling, for example, with the enhanced 3D Rotordynamics
capability, although it might take more computational time with 3D shell/solid elements to model the 3D
rotor(s), it is much easy to capture the exact geometry without special modeling constraints or efforts, which
would greatly reduce the time for model preparation, and the results are more accurate.

Main Index
 24 MSC Nastran Rotordynamics User’s Guide
Overview of MSC Nastran Rotordynamics

Main Index
 Chapter 2: Theoretical Formulation
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

2 Theoretical Formulation
 Equations of Motion 26

Equations of Motion in Fixed Reference Frame 27

Equation of Motion in Rotating Reference Frame 31

Main Index
 26 MSC Nastran Rotordynamics User’s Guide
Equations of Motion

Equations of Motion
Important features of a rotordynamic analysis can be explained using the simplified Jeffcott rotor model
introduced in the previous chapter. A front view of the same rotor is shown in Figure 2-1 and will be used in
the rest of the chapter to generate equations of motion in both the fixed and rotating coordinate systems.

Figure 2-1 Simplified Jeffcott Rotor

Assumptions:
1. Disk is rigid
2. Bar is massless
3. Linear spring support
4. Damping is neglected in the analysis
5. Rotor is spinning at constant speed Ω about the Z axis.
Equations of motions are obtained using Lagrange`s principle in both fixed and rotating coordinate systems.
The general form of Lagrange`s equation is given by:

d  ∂L  ∂L ∂D ∂W
  – + =
d t  ∂ { g· i } ∂ { g } ∂ { g· i } ∂ { g }
i i

where i = 1,2,3,…n
In the above equation, L is the Lagrange potential and is given by:
L = EK - U
where, EK is the kinetic energy, U is the potential energy, D is the dissipative energy and W is the work done
by non dissipative forces and {g} is the vector of generalized coordinates.
If the potential energy of the system is not dependent on time (true for most linear analyses), the above
expression can be simplified to:

d  ∂E K  ∂E K ∂U ∂D ∂W
  – + + =
d t  ∂ { g· i } ∂ { g } ∂ { g } ∂ { g· i } ∂ { g }
i i i
where i = 1,2,3,…n

Main Index
 Chapter 2: Theoretical Formulation 27
Equations of Motion in Fixed Reference Frame

Equations of Motion in Fixed Reference Frame

For rotating components of a model, it is important to include the effect of gyroscopic couples. The
simplified Jeffcott rotor shown in Figure 2-1 has four degrees of freedom (DOF) for lateral or transverse
vibration analysis. These DOFs are transverse displacements “u” and “v” in the X and Y direction respectively
and rotations “θ” about the X-axis and “ψ” about the Y-axis. The equations of motion are derived based on
small displacements and rotations of the rotor from its equilibrium location.
For this simplified model, the kinetic energy of the disk with respect to axes fixed in space is

1
EK 1 = m(u 2 + v 2 )
2
The kinetic energy due to the rotation of the disk is given by:q
1 1
EK 2 = I d (ω x 2 + ω y 2 ) + I pω z 2
2 2

where, ω X , ω Y and ω Z are the instantaneous angular velocities about X , Y and Z which rotate with
the disk as shown in Figure 1-2 and, Ip and Id are polar and dimeteric moment of inertias, respectively.

The instantaneous angular velocity of the disk, with respect to axes fixed in space, is θ about the X-axis and
ψ about the Y-axis.
Making use of the transformation matrices, angular velocities in a frame fixed to the rotor can be written as:

ω x   θ cos φ + ψ sin φ cos θ 

    
ω y  =  −θ sin φ + ψ cos φ cos θ 
ω   Ω − ψ sin θ 
 z  

Substituting in the above expression, the total kinetic energy is then given by,

1 1 1
EK = m ( u 2 + v 2 ) + I d (θ 2 + ψ 2 cos 2 θ ) + I d ( Ω 2 − 2 Ω ψ sin θ + ψ 2 sin 2 θ )
2 2 2

Assuming rotations are small, this expression can be simplified to:

1 1 1
EK = m ( u 2 + v 2 ) + I d (θ 2 + ψ 2 ) + I d ( Ω 2 − 2 Ω ψθ )
2 2 2

If the local coordinates are arranged in vector form { g } = {u , v , θ ,ψ } , then the inertial terms from
T

Lagrange’s equation are given by:

Main Index
 28 MSC Nastran Rotordynamics User’s Guide
Equations of Motion in Fixed Reference Frame

m 0 0 0   u  0 0 0 0   u 
0 m 0 0 0   v 
d  ∂EK  ∂EK  0   v   0 0
  − =  +Ω  
dt  ∂ { g i }  ∂ { gi }  0 0 I d 0  θ 0 0 0 − I p  θ 
    
0 0 0 I d  ψ  0 0 Ip 0  ψ 

d  ∂EK  ∂EK
 − = [ M ]{ g} + Ω [G ]{ g }
dt  ∂ { g i }  ∂ { gi }

Next, the expression for strain energy is determined based on the generalized coordinates described
previously. Since the disk and bar are assumed to be rigid in the analysis, displacements at the ends in terms
of generalized coordinates {u , v, θ ,ψ } are given as:
At end A in X direction: u − aψ

At end A in Y direction: v + aθ

At end B in X direction: u + bψ

At end B in Y direction: v − bθ
Potential energy for deformation of the springs at the end is given by:

1 1 1 1
U= K XA (u − aψ ) 2 + KYA (v + aθ ) 2 + K XB (u + bψ ) 2 + KYB (v − aψ ) 2
2 2 2 2
Contribution from potential energy to Lagrange’s equation is given by:

 K XA + K XB 0 0 − aK XA + bK XB  u 
 0 K YA + K YB aK YA − bK YB 0  
∂U
=   v 
∂ {g}  0 aK YA − bK YB a 2 K YA + b 2 K YB 0  θ 
 
 − aK XA + bK XB 0 0 a K XA + b K XB  ψ 
2 2

∂U
= [ K ]{ g }
∂ {g}

Thus, the equation of motion in a fixed reference frame without damping can be written as:

[ M ]{g} + Ω[G ]{g } + [ K ]{g} = {F }

Main Index
 Chapter 2: Theoretical Formulation 29
Equations of Motion in Fixed Reference Frame

Damping in Stator
If the supports include dampers: CXA, CYA, CXB, and CYB , then the dissipation energy is given by:

1 1 1 1
D = C XA (u − aψ )2 + CYA (v + aθ)2 + C XB (u + bψ ) 2 + CYB (v − aψ )2
2 2 2 2
Using Lagrange`s principle, contribution from dissipative energy to the structural dynamics equation of
motion is given by:

 C XA + C XB 0 0 −aC XA + bC XB   u 
 0 CYA + CYB aCYA − bCYB 0  
∂D
=    v 
∂ { g }  0 aCYA − bCYB a 2CYA + b 2CYB 0  θ 
 
 −aC XA + bC XB 0 0 a 2C XA + b 2C XB  ψ 

∂D
= [ BS ]{ g }
∂ { g }

Thus, the equation of motion in a fixed reference frame with damping defined for the stator is given by:

[ M ]{g} + Ω[G ]{g } + [ BS ]{g } + [ K ]{g} = {F }

Damping in Rotor
Damping defined for the rotating components results in damping forces proportional to displacements
defined in the rotating system. Thus, if [BR] represents the damping matrix in the rotating reference frame,
then the damping force (FDR) in the rotating coordinate system is given by:

FDR = [ BR ]{ g }

According to Chapter 1, the transformation from a fixed to a rotating frame is given by:

 cos Ωt sin Ωt 0
 0 
{v p ( t ) = [T ( Ω , t )] v p (t ) , where [T (Ω, t ) ] =  − sin Ωt
} { } cos Ωt
 0 0 1 

Thus, { g ( t )} = [T ( Ω , t )] { g ( t )} and { g (t )} = [T ] { g (t )} + Ω [T ] { g (t )}

Main Index
 30 MSC Nastran Rotordynamics User’s Guide
Equations of Motion in Fixed Reference Frame

 − sin Ω t cos Ω t 0
where  T ( Ω , t )  =  − cos Ω t − sin Ω t 0
   
 0 0 0 

Therefore, the damping force (FDF) in the fixed reference frame is given by:

[ FD F ] = [T ]
T
[ FD R ]

[ FDF ] = [T ] [ B R ] ([T ] { g ( t )} + Ω [T ] { g ( t )} )
T

[ FD F ] = [T ] [ B R ][T ] { g ( t )} + Ω [T ] [ B R ][T ] { g ( t )}
T T

If [ B R ] = bR [ I ] (where [I ] is an identity matrix), then the above expression can be simplified to:

[ F D F ] = b R { g ( t )} + b R Ω [T ] [ T ] { g ( t )}
T

[ FDF ] = bR { g (t )} + bR Ω[ J ]{ g (t )}

0 1 0
where J is a skew symmetric matrix and its given by: [ J ] =  − 1 0 0 
 0 0 0 

Thus, the rotor damping results in an additional skew symmetric force proportional to displacement. This
additional term is referred to as a circulatory matrix due to internal damping.
Equation of motion in a fixed reference frame where damping is defined for both stator and the rotor is given
by:

[ M ]{ g} + Ω [ G ]{ g } + ([ B S ] + [ B R ]){ g } + ( Ω [ K C ] R + [ K ] ) { g } = { FS }

where, [ K C ] R = b R [ J ] for this particular case.

In addition, if the rotor includes large flexible components away from the axis of rotation like disk or blades,
then the additional stiffness due to centrifugal loads needs to be included in the analysis. The modified
equation of motion in the fixed reference frame is then given by:

[ M ]{ g} + ([ BS ] + [ BR ] + Ω[G ]){ g } + ([ K ] + Ω[ KC ]R + Ω2 [ KG ]){g} = {FS }

where,
[KG] = Differential Stiffness due to radial centrifugal load

Main Index
 Chapter 2: Theoretical Formulation 31
Equation of Motion in Rotating Reference Frame

Equation of Motion in Rotating Reference Frame

As described in Chapter 1 for the analysis in a rotating reference frame, the non-rotating part of the model
must be symmetric. Hence for the model described in Figure 1-1, the following condition must be satisfied for
analysis in a rotating reference frame.
KXA = KYA = KA, CXA = CYA= CA, KXB = KYB = KB, and CXB = CYB = CB
Equations of motion in a rotating reference frame can be obtained by transforming the displacement vector.
For simplicity, consider translational DOFs only. Detailed analysis with rotational DOF is provided in the
book by Vollan and Komszik (Vollan and Komszik, 2012).
For translational DOFs, transformation from a fixed to a rotating reference frame is given by:

{ g t } = [T1 (Ω, t )]{ g t (t )}

 cos Ωt sin Ωt 
where { g t } = {u , v}T and [T1 (Ω, t )] =  .
 − sin Ωt cos Ωt 

Therefore, { g t } = [T1 (Ω , t )]T { g t + r0 } where { r0 } = { x 0 , y 0 }

T

Taking the derivative gives:

{ g t } = [T1 ]T { g t } + [T1 ]T { g t + r0 }

Kinetic energy of the disk is given by

1
E K ,t = m{g t }T {g t }
2

1
{
m [T1 ]T { g t } + [T1 ]T { g t + r0 } } {[T ] { g } + [T ] { g t + r0 }}
T
T T
E K ,t = 1 t 1
2

 − sin Ωt cos Ωt 
where, T1  = Ω 
− cos Ωt
 − sin Ωt 
The inertial terms from Lagrange’s equation are given by:

d  ∂E K  ∂E K
  − = [ M ] { gt } + 2 Ω [ C ]{ g t } − Ω 2 [ Z ] { g t } − { f cp }
dt  { i} 
∂ 
g ∂ {g i }

Main Index
 32 MSC Nastran Rotordynamics User’s Guide
Equation of Motion in Rotating Reference Frame

where

m 0
[M ] = 
m 
(mass matrix)
0

0 −m 
[C ] =  (coriolis matrix)
m 0 

m 0
[Z ] = 
m 
(centrifugal softening matrix)
0

 x0 
{ f } = mΩ
cp
2
  (centripetal force)
 y0 
The centripetal force acting on the rotor disk is proportional to the square of the operating speed and distance
from the axis of rotation. If higher-order strain is included in the analysis, an axial load results in additional
bending stiffness, which is referred to as geometric or differential stiffness. It is denoted by: Ω2[KG]
Since the support stiffness is the same in both directions, the simplified stiffness matrix is the same in both
the fixed and rotating systems. Therefore, the contribution from strain energy for analysis performed in a
rotating coordinate system is given by:

∂U K + KB 0 
= A + Ω2[KG ]
∂ {g}  0 K A + K B 

Damping in Rotors
Damping defined for rotors can be expressed as (as described earlier):

FDR = [ B R ]{ g }

Thus, the equation of motion with damping defined for rotors is given by:

[ M ] { gt } + 2 Ω [C ]{ g t } + [ B R ] { g t } + ([ K ] + Ω 2 [ K G ] − Ω 2 [ Z ] ) { g t } = { FR }

Damping in Stator
As discussed earlier, damping defined in the fixed reference frame for the stator needs to be converted into
damping defined in the rotating reference frame using coordinate transformation.

Therefore, the damping force in the fixed reference frame for stator is given by:

Main Index
 Chapter 2: Theoretical Formulation 33
Final Equations of Motion

FDS = b { g t } , where b = CA + C B.
Transforming this damping force to the rotating reference frame:

(
FDS = b[T1 ] [T1 ]T { g t } + [T1 ]T { g t } )

( )
FDR = b[T1 ] [T1 ]T { g t } + [T1 ]T { g t + r0 } = b { g t } + Ω b[ J t ] { g t }

FDS = [ B S ] { g t } + Ω [ K C ] S { g t }

where, [BS] = b[I] and [KC] S = b[Jt].

Thus, the equation of motion in the rotating reference frame with damping defined for both the rotor and
the stator is given by:

[ M ] { gt } + ( 2 Ω [C ] + ([ B R ] + [ BS ]) ) { g t } + ([ K ] + Ω 2 [ K G ] − Ω 2 [ Z ] + Ω[ K C ]S ) { g t } = { FR }

Final Equations of Motion

The general equation of motion for a model with rotating components is given by:
1. Final equation of motion in a rotating reference frame:

[ M ]{ g} + ([ BS ] + [ BR ] + 2Ω[C ] ) { g } + ([ K ] − Ω 2 [ Z ] + Ω 2 [ K G ] + Ω[ K C ]S ) { g } = { FR }
(2-1)
where,
[M] = Mass matrix

{g} = Degrees of freedom in the rotating reference frame

[BS] = Damping in fixed frame
[BR] = Damping in rotating frame
Ω= Operating rotor speed
[C] = Coriolis Effect
[K] = Stiffness matrix
[Z] = Centrifugal softening matrix
[KG] = Differential Stiffness due to axial centrifugal load
[KC]S = Circulatory matrix due to damping in fixed frame
{FR} = External force vector in rotating reference frame
2. Final equation of motion in a fixed reference frame:

Main Index
 34 MSC Nastran Rotordynamics User’s Guide
Final Equations of Motion

[ M ]{ g} + ([ BS ] + [ BR ] + Ω[G ]){ g } + ([ K ] + Ω[ K C ]R + Ω 2 [ K G ]) {g} = {FS }

(2-2)
where,

{ g } = Degrees of freedom in the fixed reference frame

[G] = Gyroscopic Matrix
[KC]R = Circulatory matrix due to damping in rotating frame
{FS} = External force vector in fixed reference frame

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Elements Supported for

3 Rotordynamics Analysis


Rotor Modeling 36

Main Index
 36 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Rotor Modeling

Rotordynamic Analysis using MSC Nastran

MSC Nastran supports rotordynamic analysis for 1D, 2D axisymmetric harmonic and 3D rotor geometries.
The three corresponding cards and their element support are listed in Table 3-1. It should be noted that all
three cards are valid for analysis in the fixed reference frame and that the ROTOR card furthermore supports
analysis that takes place in the rotating reference frame.
After the ROTORG card was released in MSC Nastran 2004, the new 2D and 3D capabilities were
introduced in MSC Nastran 2014 and MSC Nastran 2017, respectively. In addition to direct solution
approaches, modal solution approaches can also be used for complex eigenvalue analysis (SOL 110) and
complex frequency response analysis (SOL 111) for all the rotors. Note that modal transient analysis (SOL
112) is not supported for rotordynamic analysis.

Table 3-1 Rotor definition entries

Rotor Entry Reference Frame
Name Geometric Extent Elements Supported Supported
ROTORG 1D CONM2, CBAR, CBEAM Fixed only
ROTORAX 2D Axisymmetric CONM2, CQAUDX, CTRIAX Fixed only
Harmonic
ROTOR 3D CONM1,CONM2 Both fixed and rotating

CBAR, CBEAM

CQUAD4, CQUAD8, CTRIA3,

CTRIA6

CTETRA, CPENTA, CHEXA

A brief preview of each of these rotor cards is given in Figure 3-1 , Figure 3-2, and Figure 3-3. Detailed
descriptions are available in the MSC Nastran Quick Reference Guide.

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 37
Rotor Modeling

ROTORG Rotor Line Model Grids

Specifies grids that compose the rotor line model.

Format:
1 2 3 4 5 6 7 8 9 10
ROTOR ROTORID GRID1 GRID2 ... GRIDn
G

Figure 3-1 ROTORG Definition

ROTORAX Axisymmetric Model Rotor Definition

Specifies list of elements, properties or grid points that comprise an axisymmetric model rotor.

Format:
1 2 3 4 5 6 7 8 9 10
ROTORAX ROTORID LTYPE ID1 ID2 ... ... IDn

Figure 3-2 ROTORAX Definition

ROTOR Rotor Model Definition

Specifies list of grids, elements or properties that comprise the rotor 3D model.

Format:
1 2 3 4 5 6 7 8 9 10
ROTOR ROTORID FRAME
LTYPE ID1 ID2 ID3 etc.
AXIS GID1 GID2 etc.

Figure 3-3 ROTOR Definition

In order to define a rotor model, the following information needs to be provided:

a. Unique rotor identification number
b. List of elements/nodes that comprise the rotor model
c. At least two grid points to define the axis of rotation

Main Index
 38 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

d. Type of reference frame to use for analysis

In the case of ROTORG, the list of nodes specified through GRID1, GRID2, … GRIDn are sufficient to
define the components of the rotor. Since these grid points must be collinear, they also define the axis of
rotation. (In case these points are not collinear, a FATAL message is issued by MSC Nastran). It is important
to realize that ROTORG grids may not be connected directly to any element that is not part of the rotor.
However, since physical rotor/stator couplings via bearings commonly occur in rotor systems, special
modeling techniques are available to specify such connections.
In the case of ROTORAX, components of the rotor can be specified either by listing all the elements that
comprise the rotor (LTYPE = ELEM) or by specifying property IDs associated with the elements that
comprise the rotor (LTYPE = PROP). The axis of rotation is specified using LTYPE = GRID. At least two
grid points are required to specify the axis of rotation. These grid points need not be part of the axisymmetric
harmonic rotor.
ROTOR is similar to ROTORAX but more condensed in that all rotor details for a single rotor can be
specified in one single card. Unlike the previous two rotor entries, here the user needs to specify the reference
frame being used for the analysis next to the rotor ID. In a single test deck, only one type of reference frame
can be used for analysis, either rotating or fixed.

Sample Test Case

Figure 3-4 Uniform rigid rotor supported at ends

To illustrate this Rotordynamics functionality within MSC Nastran, let us consider the hollow rotor with
two bearing supports in Figure 3-4 and model it with the introduced ROTORG, ROTORAX and ROTOR
cards in sequence. The rotor has a length of 0.5m, an inner diameter of 0.1m, and an outer diameter of 0.2m.
Both bearing supports have horizontal and vertical stiffnesses of 1MN/m. The rotor is made of steel with a
density of 7810 kg/m3 and an elastic constant of 210 GPa.

Model with ROTORG

The ROTORG entry can be used to define a line rotor by specifying grid points belonging to the rotor. A
rotor defined using the ROTORG entry may consist of CBAR, CBEAM, and/or CONM2 elements. It
should be noted that the CBAR elements do not include a polar moment of inertia in the element
formulation. Hence, CBAR elements (with default settings) may be used in combination with CONM2
elements to generate gyroscopic matrices for analysis in a fixed reference frame. Although the CONM2 mass

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 39
Rotor Modeling

elements can also be optionally associated with CBEAM elements, the PBEAM card (i.e.: CBEAM property
card) allows for the direct specification of a polar moment of inertia.
Rotors can be firmly constrained in space with SPCs or mounted with bearings supports using rigid elements
(e.g. RBAR, RBE2), interpolation elements (e.g.: RBE3) or constraint equations (e.g. MPC). For more detail
about these special elements, the user can refer to the MSC Nastran Quick Reference Guide (QRG).If
connection using conventional elements between the rotor and bearing or the support structure is detected,
a fatal error will be issued. It is important to have the rotor detached from the support structure because the
inertial properties for the calculation of the gyroscopic terms are determined using the G-size matrices.
Connections to ground or support elements may cause incorrect determination of the gyroscopic terms. For
the example shown here, the end points of the rotor are connected to ground with springs through RBE2
elements.
a. ROTORG using CBEAM elements
The rotor modeled using four CBEAM elements is shown in Figure 3-5 and the test deck for the rotor part of
this model is shown in Figure 3-6. The rotor with ID 1 is defined as consisting of nodes 1 to 5 using the
ROTORG entry. Structural properties and shape for the beam are defined using MAT1 and PBEAML
entries, respectively. In the current example, bearings are defined in the fixed reference frame. In order to
make that distinction, two grid points (Node 11 and 15) are defined coincident with existing rotor nodes.
Node 11 is connected to Node 1 and Node 15 is connected to Node 5 using RBE2 elements. Finally, spring
supports are defined on nodes 11 and 15.

Figure 3-5 Rotor model with CBEAM entries

For the rotor model defined using CBEAM elements as shown in Figure 3-5 and Figure 3-6, the corresponding
rotor summary in the .F06 is shown in Figure 3-7. For rotors defined using a ROTORG entry, the following
information about the rotor is provided in the output.F06 file:
a. Grid points found on each rotor.
b. Assumed direction of rotation based on the grids defined in the ROTORG entry. (The final spin
direction for each rotor is provided by the RSPINR entry.)
c. Rotor mass summary which includes details for inertia properties for each of the grid points that
comprise the rotor model. Since a CBEAM element is used in the analysis, the mass summary for
each grid point also includes polar moment of inertia. The last line in this summary shows the
Center of gravity for the rotor model.

Main Index
 40 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Figure 3-6 Test deck for rotor model with CBEAM elements

Figure 3-7 Rotor Summary in the.F06 file for the rotor defined using CBEAM element

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 41
Rotor Modeling

d. ROTORG using CBAR and CONM2 elements

If a massless shaft is desired in a Rotordynamics analysis, the entire mass and inertia of the shaft can be
assigned to a concentrated mass located at the center. This particular case can be modeled using CBAR and
CONM2 elements as shown in Figure 3-8 and Figure 3-9.

Figure 3-8 Rotor model with CBAR and CONM2 elements

Figure 3-9 Input deck for rotor model with CBAR and CONM2 elements

With reference to the test deck shown in Figure 3-6, the following changes are made in the model:
 PBEAML is replaced by PBARL
 CBEAM is replaced by CBAR
 The material density in the MAT1 card is changed from 7810.0 to 0.0 kg/m3.
 A concentrated mass is added at the center of the rotor model.

Main Index
 42 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Figure 3-10 Results for the rotor model with CBAR/CONM2 elements

The only difference in the results presented in Figure 3-7 and Figure 3-10 is in the rotor mass summary. As
expected for the model with CBAR and CONM2 elements, the entire mass and inertia of the model is located
at the center of the rotor on the grid with ID 3. Of course, the entire mass does not have to be concentrated
in one GRID point and can be distributed over several or even all of them.

Model using ROTORAX

The hollow rotor modeled using axisymmetric harmonic elements is shown in Figure 3-11. In general, the
number of axisymmetric elements required to obtain a high fidelity model is much larger than the number
of elements required for a beam/bar element model. However, for this particular example, in order to
highlight the usage of axisymmetric elements, only four CQUADX elements are used in the analysis.

Figure 3-11 Rotor model developed using axisymmetric harmonic elements

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 43
Rotor Modeling

Figure 3-12 Test deck for rotor model with axisymmetric harmonic elements

Part of the test deck for the rotor modeled using CQUADX elements is shown in Figure 3-12. The rotor with
ID 1 is defined by specifying the property of the elements that comprise the rotor and the grid points that
define the axis of rotation. For the case of ROTORAX entries, the PROP entry must point to a PAXSYMH
entry. For this particular case, the ELEM option could have also been used to define the components of the
rotor as shown below:
$ ROTOR DEFINITION
$ ROTORAX 10 PROP 1
ROTORAX 10 ELEM 1 THRU 4

The details of the PAXSYMH card used for specifying the properties of the elements is described in the MSC
Nastran Quick Reference Guide and Appendix. Note that in the current analysis, a harmonic index of 1 is used in
order to perform lateral bending analysis. For axial and torsion analysis, the harmonic index needs to be
changed from “1” to “0”.
The CQUADX and GRID entries are defined in the conventional way. In order to connect the rotor to the
conventional springs, two “3-D” grid points are defined. (A grid point associated with an axisymmetric
harmonic element is considered to be an axisymmetric harmonic grid point. As described in Appendix B of

Main Index
 44 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

this manual, the axisymmetric harmonic grid points have six DOFs, and they are different from conventional
3-D grid points.) In order to make a connection between the axisymmetric harmonic grid point and the 3-
D grid, the RBAX3D connector elements are used. In the current model, two RBAX3D elements are used,
one at either end, to connect the rotor to the bearings at the ends. Once the connection to 3-D grid points
has been made, CBUSH elements can be attached to these 3-D grid points for the support. Rules related to
the usage of RBAX3D entry are given in the MSC Nastran Quick Reference Guide and Appendix B of this
manual.

Figure 3-13 Rotordynamics summary from axisymmetric harmonic rotor

The rotordynamics summary in the .F06 file for a rotor model defined using ROTORAX entry is shown in
Figure 3-13. It lists the elements defined on the axis of rotation. Any 3-D grid point attached to an
axisymmetric harmonic rotor using the RBAX3D entry is also considered part of the rotor axis grid list. The
rotor spin direction is determined based on the grid points specified under the GRID option. And finally, the
rotor mass summary for each of the rotors defined in the model is listed. It includes total rotor mass, its CG
location and moment of inertia. It can be seen here that the total mass, CG location, and polar moment of
inertia for the beam rotor and axisymmetric rotor are the same.
In some cases, it may be desirable to treat the rotor as massless and assign mass and inertia properties to a
concentrated point mass, as simulated earlier using CBAR/CONM2 elements. This can also be done using
axisymmetric harmonic elements as shown in Figure 3-14. Here an additional GRID with ID 13 is defined on
the axis of symmetry at the center of the rotor (i.e., where the point mass needs to be defined). This 3D GRID
point is connected to a point on the axisymmetric rotor using RBAX3D entry. Now, a point mass can be
defined on this 3-D grid point which will be treated as part of the rotor during the analysis. Results for this
particular case are shown in Figure 3-15. Now the Rotordynamics summary shows three points listed on the
rotor axis compared with two points listed in Figure 3-13. Also, the mass summary for each rotor includes an
additional line which lists the total mass of the axisymmetric rotor after including the additional mass defined
using the CONM2 element.

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 45
Rotor Modeling

Figure 3-14 Test deck with axisymmetric harmonic elements and CONM2

Figure 3-15 Results for rotor defined using axisymmetric harmonic elements and CONM2

Main Index
 46 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Model with ROTOR

a. Using shell elements
Finally, the rotor is modeled using shell elements CQUAD4 and the new ROTOR entry. The rotor model
and the corresponding test deck are shown in Figure 3-16 and Figure 3-17, respectively.

Figure 3-16 Rotor model with shell elements

Figure 3-17 Test deck with shell elements for analysis in fixed system

The ROTOR entry includes all the details about the reference frame for analysis (FIX), components of the
rotor (PROP 1) and grid points to define the axis of rotation (AXIS 220 221). The material properties for

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 47
Rotor Modeling

the shell elements are defined using usual PSHELL and MAT1 entries. As shown above, RBE2 elements
(RBE3 element can also be used) are used here to connect the rotor to the bearings in the fixed frame. Also,
it can be easily seen here that the size of the model has increased by at least 40 times compared with the beam
rotor model.
The rotordynamics summary for the rotor model defined using shell elements is shown in Figure 3-18. It
follows the same format as that for the axisymmetric rotor. The .F06 file lists the grid points that lie on the
axis of rotation and define the rotor axis. The total mass and the rotor CG location obtained for the rotor
with shell elements is the same as that obtained earlier for the rotor with beam elements and the rotor with
axisymmetric harmonic elements. Note that the polar moment of inertia is zero in this case since the
rotational degrees of freedom do not include inertia effects for shell elements.

Figure 3-18 Rotor summary for the 3-D rotor

b. Using ROTOR entry for 1-D elements

As described in Table 3-1, the rotor defined using CBEAM, CBAR and CONM2 elements can also be defined
using the ROTOR entry. Thus, the rotor defined earlier in Figure 3-5 and Figure 3-8 can also be defined using
the new ROTOR entry.
The 1-D rotor with beam elements can be defined using the ROTOR entry as:
$ ROTORG 1 1 THRU 5
ROTOR 1 FIX
PROP 1
AXIS 1 2

Main Index
 48 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Similarly, the 1-D rotor modeled using CBAR and CONM2 element can be defined using the ROTOR entry
as:
$ ROTOR DEFINITION
$ ROTORG 1 1 THRU 5
ROTOR 1 FIX
ELEM 21
PROP 1
AXIS 1 2

Since there is no property associated with the CONM2 elements, it is defined as part of the ROTOR entry
using the ELEM option, as observed above. The Rotordynamics summary for the above case is shown in
Figure 3-19. The rotor information provided here is similar to that listed in Figure 3-10.

Figure 3-19 Rotordynamic summary for ROTOR with CBAR/CONM2 elements

Rotor Modeling Guidelines

In this section, a brief overview of different rotor modeling approaches is provided. In general, the approach
used in the analysis depends on the following features of the model:
1. Symmetry present in rotor and stator
2. Complexity of the rotor
3. Computational time
The symmetry present in rotor and stators identifies the reference frame to be used for the rotordynamic
analysis. The next important criteria is to determine the complexity of the rotor model and the best modeling
technique to use to capture the geometry of the model. Even though higher-order elements may capture the
model details better, it comes at the expense of significant computational time. (In some cases, the use of
enhanced reduction techniques can offset the penalty associated with higher order models. These reduction
techniques are discussed in a later chapter.) Here, four representative rotor models are presented, and for each

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 49
Rotor Modeling

of the case, appropriate modeling recommendation is made to obtain accurate results without significant
computational time.

Case 1: Uniform Hollow Rotor Model

A uniform hollow rotor modeled using three different options available for analysis in fixed reference frame
is shown in Figure 3-20. Number of elements and grid points used for generating these rotor models is listed
in Table 3-2. Here it can be seen that, as compared to a one-dimensional rotor model, the two-dimensional
rotor model requires 90 times more nodes whereas the three-dimensional model requires 290 times nodes.
Thus for this simple and uniform hollow rotor, it is sensible to use one-dimensional rotor modeled using
CBEAM (or CBAR and CONM2) and defined using ROTORG to save computational time.

Figure 3-20 Hollow Rotor Model

Table 3-2 Finite element details

Model # of Grids # of Elements
ROTORG 10 9
ROTORAX 904 675
ROTOR 2900 2870

Case 2: Symmetric Nonuniform Rotor Model

Figure 3-21 shows the Nelson-MacVaugh (Nelson and Vaugh, 1976) rotor modeled using different elements.
As seen in the figure, the rotor is nonuniform and diameter varies across the length of rotor.

Main Index
 50 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Figure 3-21 Nelson-MacVaugh (NM) Rotor

For this case, the one-dimensional rotor modeled using CBEAM elements may not capture the stiffness of
the rotor model correctly in the region where the rotor diameters varies sharply. For these kind of models, the
use of axisymmetric harmonic elements, as shown in Figure 3-21(b), is recommended. In this case, the 3-D
solid elements (Figure 3-21(c)) also captures the changes in rotor geometry correctly; however, they require
much more nodes and elements to capture the correct geometry, and thus they are computationally more
expensive.

Case 3: Symmetric Nonuniform Rotor

Model for a rotor shaft with propeller is shown in Figure 3-22. The finite element model consists of CQUAD4
and solid CHEXA elements. This particular model cannot be analyzed accurately using one-dimensional or
two-dimensional elements. Since the rotor has discrete blades attached to be, the rotor model shown here is
not axisymmetric by definition. However, as the rotor blades are uniformly distributed around the
circumference, the rotor model can be approximately treated as symmetric and analyzed in fixed reference
frame. Thus, this model needs to be defined using the ROTOR entry, either in fixed or stationary reference
frame, depending on the nature of support structure. (If the discrete blades in the model are short and can be
treated as rigid, then all the blades in the rotor model can be replaced by a concentrated mass (CONM2)
element at the center of the axis of rotation/symmetry. In this case, it would be possible to model the rotor
using axisymmetric harmonic elements and define is using ROTORAX entry.)

Main Index
 Chapter 3: Elements Supported for Rotordynamics Analysis 51
Rotor Modeling

Figure 3-22 Propeller with shaft

Case 4: Nonsymmetric Rotor

The model for a crankshaft generated using solid CHEXA and CPENTA elements is shown in Figure 3-23.
This model has nonsymmetric rotating structure. Thus this model can ONLY be analyzed using the
ROTOR entry in rotating reference frame, assuming it has symmetric boundary conditions and it is attached
to symmetric supports.

Figure 3-23 Crankshaft model generated using solid elements

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 3-3
Input File Name Description
rug_3_1 Rotor with CBEAM elements.
rug_3_2 Rotor with CBAR and CONM2.
rug_3_3 Rotor with axisymmetric elements.
rug_3_4 Rotor with axisymmetric elements and CONM2.
rug_3_5 3D rotor with shell elements.

Main Index
 52 MSC Nastran Rotordynamics User’s Guide
Rotor Modeling

Main Index
 Chapter 4: Damping
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

4 Damping


Damping and Bearing Models 54

Linear Bearings 59
 Nonlinear Bearings 61

Main Index
 54 MSC Nastran Rotordynamics User’s Guide
Damping and Bearing Models

Damping and Bearing Models

Several damping options are available in MSC Nastran for both the rotor and stator components. Due to the
presence of circulatory terms, damping for rotors is separated and treated differently from damping for the
stator portion of the model. Besides applying damping directly to the structural model, the user can also
include linear/nonlinear dampers in the model using Bulk Data entries like CDAMP, CBUSH, etc. In order
to model complex behavior associated with bearings in rotordynamics, several models are available such as
the two-dimensional linear/nonlinear bush element (CBUSH2D), nonlinear squeeze film damper
(NLRSFD), nonlinear radial load (proportional to gap, NLRGAP), and the nonlinear load as a tabular
function (NOLINi).

Structural Damping Options for Stator

The following options are available in MSC Nastran for applying damping to structural model of the stator.
(For each of the available options, Bulk data entries that can be used to specify the damping parameters, and
mathematical expressions used for applying the damping, are also listed.)

Structural Damping
Sources: Param G or Field ‘G’ in DAMPING card
Application:

1. Added to stiffness matrix: [KS]G =

(Only for frequency response and complex eigenvalue solutions)

2. Added to Damping matrix: [BS]G = (W3 from PARAM, W3)

where, [KGG] is G-size stiffness matrix for the stator and W3 ≠ 0. (Only in transient solutions)

Material Damping
Sources: Field ‘GE’ in material property card (eg. MAT1)
Application:

1. Added to stiffness matrix: [K4s] = (for each material type)

(Only for frequency response and complex eigenvalue solutions)

2. Added to Damping matrix: [BS]4 = (W4 from PARAM, W4)

where W4 ≠ 0. (Only in transient solutions)

Rayleigh Damping
Sources: Field ‘Alpha1’ and ‘Alpha2’ in DAMPING card or PARAM, ALPHA1 and PARAM ALPHA2.

Main Index
 Chapter 4: Damping 55
Damping and Bearing Models

Application:

1. Added to Damping matrix: [BS] =

where, [MGG] is G-size mass matrix for the stator.

Hybrid Damping
Sources: Field ‘HYBRID’ in DAMPING card and HYBDAMP card.
Application:
Case 1: KDAMP = No

1. Added to Damping matrix: [BS]H =

Case 2: KDAMP = Yes
1. Added to Stiffness matrix: [KS]H = i[KH]
(Only for frequency response and complex eigenvalue solutions)

2. Added to Damping matrix: [BS]H = [-------------

KH ]
(WH from DAMPING entry)
WH

where WH ≠ 0 in transient solutions.

(Refer to QRG for further details related to matrices [BH] and [KH] under the description for bulk
data entry HYBDAMP)

Modal Damping
Sources: SDAMPING in CASE Control and TABDMP1 or TABLEDi in Bulk Data section.
Application:
1. Added to Damping matrix: [BS]
Diagonal terms of the damping matrix are given by:

bi = ξiωi mi

where, ωi is the undamped vibration frequency determined using [M] and [K] matrices and ξi is
the dimensionless input coefficient.
Note that the modal damping matrix generated by this approach is not affected by the presence of rotors in
the model. Also, the circulation terms are not generated for damping corresponding to rotor degrees of
freedom.

Main Index
 56 MSC Nastran Rotordynamics User’s Guide
Damping and Bearing Models

Damping Options for Rotating Structure

The damping options described previously for the stator are also available for the rotating part of the
structure. However, they need to be specified through the RSPINR/RSPINT entry which is defined for each
rotor in the model. As described previously, the presence of damping terms in rotating systems leads to a
stiffness-like circulatory matrix in the equations of motion that is proportional to rotor speed.
In the case of material damping, structural damping and hybrid damping with “KDAMP=YES” for rotors,
it is possible to add the damping term either to the [B] matrix or to the [K] matrix. The actual
implementation depends upon the type of analysis (complex eigenvalue, frequency response or transient
response) and the presence of parameters WR3, WR4, and WRH. In addition, these parameters can also be
defined individually for each rotor using RSPINR/RSPINT entries. (Here, they are referred to as WR3R,
WR4R and WRHR, respectively.)

Structural Damping
Sources: Field ‘GR’ in RSPINR/RSPINT card
Application:
1. Added to stiffness matrix: [KR]GR =
(Only for frequency response when PARAM, GYROAVG ≥ 0)

2. Added to damping matrix: [BR]GR = (WR3 from PARAM, WR3)

where [KRGG] is the G-size stiffness matrix for the rotor and WR3 ≠ 0.
(For complex eigenvalue analysis, frequency response when PARAM, GYROAVG < 0 and transient
analysis)

Material Damping
Sources: Field ‘GE’ in material property card
Application:

1. Added to stiffness matrix: [K4R] = (for each material type)

(Only for frequency response when PARAM, GYROAVG ≥ 0)

2. Added to damping matrix: [B4R] = (WR4 from PARAM, WR4)

where WR4 ≠ 0.
(For complex eigenvalue analysis, frequency response when PARAM, GYROAVG < 0 and transient
analysis)

Rayleigh Damping
Sources: Field ‘AlphaR1’ and ‘AlphaR2’ in RSPINR/RSPINT card
Application:

Main Index
 Chapter 4: Damping 57
Damping and Bearing Models

1. Added to damping matrix: [BR] =

where [MRGG] is the G-size mass matrix for the rotor.

Hybrid Damping
Sources: ROTHYBD (separate bulk data entry and not in RSPINR/RSPINT) and HYBDAMP card.
Application:
Case 1: KDAMP = No
1. Added to damping matrix: [BR]H = [BHR]
Case 2: KDAMP = Yes
1. Added to stiffness matrix: [KR]H = i[KHR]
(Only for frequency response when PARAM, GYROAVG ≥ 0)
[ KH ]
R
2. Added to damping matrix: [BR]H = ---------------- (WRH from PARAM, WRH)
WRH

where [KRGG] is the G-size stiffness matrix for the rotor and WRH ≠ 0.
(For complex eigenvalue analysis, frequency response when PARAM, GYROAVG < 0 and transient
analysis)

Circulation Matrix
As described in Chapter 2, the presence of rotor damping when the problem is analyzed in a fixed system
results in a circulatory matrix ( [ KC ]R ). This matrix adds to the system stiffness matrix. It is multiplied to
operating rotor speed and is skew symmetric in nature.

If the damping force for the rotor is given by the following expression: FDR = cR { g } , then the circulation
matrix is given by: [ K C ]R , B = cR [ J ] where [J] is a skew symmetric matrix.

For a more general case, the circulatory matrix corresponding to each 3-D grid “j” is given by:

[ K C ]R , B = [TI ][ BR ] j + [ BR ] j [TI ]

where,
[BR]j = Viscous damping matrix corresponding to grid point “j”,
[TI] = skew symmetric matrix with the following form in the rotor coordinate system, and

Main Index
 58 MSC Nastran Rotordynamics User’s Guide
Damping and Bearing Models

 0 0.5 0 0 0 0
 −0.5 0 0 0 0 0 

 0 0 0 0 0 0
[TI ] j = 
 0 0 0 0 0.5 0 
 0 0 0 −0.5 0 0 
 
 0 0 0 0 0 0

(The above matrix is only valid for 3-D grid points; that is, for rotors defined using ROTORG and ROTOR
entries. A different matrix is used for axisymmetric harmonic grid points where rotors are defined using
ROTORAX.) The 6x6 [TI]j matrix is repeated for each grid point that is part of the rotor.
Thus, the circulation matrix for different damping options defined is given by:
a. Viscous dampers in Rotor [BR] (described in the next section)
Corresponding circulation matrix: [ K C ]R , B = [TI ][ B R ] + [ B R ][TI ]
b. Structural damping GR
GR
Corresponding circulation matrix: [ KC ]R ,GR =  K RC  (if WR3 ≠ 0)
WR3  

where  K R  = [TI ][ KRGG ] + [ KRGG ][TI ]

C

c. Material damping
1
Corresponding circulation matrix: [ K C ]R , GE =  K 4 CR  (if WR4 ≠ 0)
WR 4 

where  K 4 R  = [TI ][ K 4 RGG ] + [ K 4 RGG ][TI ]

C

d. Rayleigh damping
Corresponding circulation matrix: [ K C ]R ,α = α 1 R  M R  + α 2 R  K R 
C C

where  M R  = [TI ][ MRGG ] + [ MRGG ][TI ]

C

e. Hybrid damping
• If KDAMP = YES
1
Corresponding circulation matrix: [ K C ]R , KH =  KH RC  (if WRH ≠ 0)
WRH 
where  KH R  = [TI ][ KHRGG ] + [ KHRGG ][TI ]
C

• If KDAMP = NO
Corresponding circulation matrix: [ K C ]R , BH = [ B H RC ] = [T I ][ B H R ] + [ B H R ][T I ]
Similar expressions are also obtained when the problem is solved in a rotating coordinate system and viscous
damping (proportional to velocity) is defined in the fixed coordinate system.

Main Index
 Chapter 4: Damping 59
Linear Bearings

Effect of Circulation Terms

Often for the analysis in a fixed reference frame, the damping defined for the rotor is referred to as the internal
damping. Internal damping is primarily due to friction at the rotor component interfaces, and it can be
modeled using any of the damping models described above. When the power inserted by internal damping
exceeds the power extracted by external damping, the rotor will become unstable. For this instability to
manifest itself the rotor must be operating above a critical speed whose mode shape promotes interfacial
friction and it must be forward whirl (Nelson, 2007).

Controlling Rotor Damping and Circulation

For some practical applications, it is desirable not to include the effect of circulation matrices. In several cases,
it may be cumbersome to define rotor damping separately from the rest of the model. In order to
accommodate these practical scenarios, a new parameter called “RDBOTH(695) ” has been introduced using the
Bulk Data entry MDLPRM. This parameter can be define in the Bulk data using the following format:
MDLPRM RDBOTH N
where N = 0 (default),1 or 2. Changes made in the analysis for different values of N are described below:

RDBOTH Parameter to select Rayleigh damping approach as implemented in V2005, Integer. A

cumulative sum can be provided in case more than one feature are desired in the analysis.
0 Uses current implementation for Rayleigh Damping as described in
RSPINR/RSPINT entry description (Default)
1 Switch to V2005 implementation of Rayleigh damping where damping coefficients
specified in the model through “PARAM, ALPHA1” and “PARAM, ALPHA2” are
applied to the complete model and Rayleigh damping specified through
“ALPHAR1” and “ALPHAR2” in RSPINR/RSPINT is set to 0.0.
2 Ignore circulation effects in rotordynamic analysis.

Linear Bearings
Connections between rotor and stator (to simulate bearings) or from rotor to rotor (to simulate gears) can be
made using scalar elements, like CELASi and CDAMPi, and bushing element CBUSH.

Scalar Elements
Scalar elements are connected between pairs of degrees of freedom (at either scalar or geometric grid points)
or between one degree of freedom and ground. Scalar elements are available as springs and viscous dampers.
Scalar spring elements (CELASi) are useful for representing elastic properties that cannot be conveniently
modeled with the usual structural elements (elements whose stiffnesses are derived from geometric
properties). The scalar damper (CDAMPi) is used to provide viscous damping between two selected degrees
of freedom or between one degree of freedom and ground.

Main Index
 60 MSC Nastran Rotordynamics User’s Guide
Linear Bearings

The most general definition of a scalar spring is given with a CELAS1 entry. The associated properties are
given on the PELAS entry. The properties include the magnitude of the elastic spring, a damping coefficient,
and a stress coefficient to be used in stress recovery. The CELAS2 defines a scalar spring without reference to
a property entry. The CELAS3 entry defines a scalar spring that is connected only to scalar points, and the
properties are given on a PELAS entry. The CELAS4 entry defines a scalar spring that is connected only to
scalar points and without reference to a property entry. No damping coefficient or stress coefficient is
available with the CELAS4 entry. Element force is calculated from the equation:

F = k (u1 − u2 )

where “k” is the stiffness coefficient for the scalar element and “u1”is the displacement of degree of freedom
i listed on its connection entry. Element stresses are calculated from the equation

σ = S .F
where “S” is the stress coefficient on the connection or property entry and is as defined above.
Grounded connections are indicated on the connection entry by leaving the appropriate scalar identification
number blank. The CDAMP1, CDAMP2, CDAMP3, CDAMP4, and CDAMP5 entries define scalar
dampers in a manner similar to the scalar spring definitions. The associated PDAMP entry contains only a
value for the scalar damper.

BUSH Elements
The generalized spring-damper element CBUSH is a structural scalar element connecting two noncoincident
grid points, or two coincident grid points, or one grid point to ground using an associated PBUSH entry for
properties. This combination is valid for any structural solution sequence. To make the PBUSH frequency
dependent it need only have an associated PBUSHT Bulk Data entry. The PBUSHT entry, for frequency
dependency, is only used in SOL 108 and SOL 111. The stiffness matrix for the CBUSH element takes the
diagonal form in the element system.

Cross Coupling in Bearing Supports

Consider the example where a grid with ID 91 is connected to ground using a 2-D spring which has
cross-stiffness terms. Cross coupling terms for modeling bearings can be introduced in the analysis using one
of the three approaches:

Figure 4-1 Model with cross-coupled spring

Main Index
 Chapter 4: Damping 61
Nonlinear Bearings

a. CROSS Option in PBUSH2D (introduced in V2018)

Format:
1 2 3 4 5 6 7 8
PBUSH2D PID K11 K22 B11 B22 M11 M22
CROSS K12 K21 B12 B21 M12 M21

CBUSH2D 201 211 91 1092 0 XY

PBUSH2D 211 3.503+7 3.503+7
CROSS 8.756+6 8.756+6

Here, grid 1092 is connected to ground.

b. DMIG approach
DMIG STIF 0 1 1
DMIG STIF 91 1 91 1 3.503+7
91 2 8.756+6
DMIG STIF 91 2 91 1 8.756+6
91 2 3.503+7
c. Rotated CBUSH approach (can only be used for symmetric matrices)
CBUSH 202 211 91
PBUSH 211 K 2.627+7 2.627+7
CORD2R 1 0 0. 0. 0. 0. 0. 1.
1. 1. 0.
CBUSH 204 212 91
PBUSH 212 K 1.751+7

Nonlinear Bearings
Currently, MSC Nastran does not support nonlinear analysis for Rotordynamics where the nonlinearity is
related to material property, geometry, and contact. However, in transient response, it is possible to include
the effect of nonlinear bearing forces. Nonlinear bearings that can be used for rotordynamic analysis are
briefly described here. Further details are available in MSC Nastran Dynamic Analysis User’s Guide .
Nonlinear bearings are basically elastic elements with nonlinear force-deflection relationships. Some of these
elements also exhibit nonlinearities proportional to the velocity. For the elements described here, the pseudo-
force method is used in MSC Nastran where the deviations from linearity are treated as additional applied
loads. The dynamic equations of motion can be written as:

[ M ]{ g(t )} + [ B ]{ g (t )} + [ K ]{ g (t )} = {P (t )} + { N (t )}
where, [M], [B] and [K] are system mass, damping and stiffness matrices respectively. Note that the [B] and
[K] matrices shown here include all the rotordynamic related terms. The vectors {P(t)} and {g(t)} denote

Main Index
 62 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

applied nodal loads (linear) and system displacements respectively, as functions of time. The vector {N(t)}
denotes nonlinear force produced by bearing elements.

Nonlinear Gap, NLRGAP

The NLRGAP entry allows you to model contact between two coaxial cylinders; for example, a shaft and its
housing.
The description of the NLRGAP bulk data entry is shown in Figure 4-2. The NLRGAP is an idealization of
the contact between any circular shaft enclosed by a circular housing (Figure 4-3). In this idealization, grid
point A represents the outer surface of the shaft while grid point B represents the inner surface of the housing.
Grid points A and B are required to be coincident and must have a parallel displacement coordinate system.
A rudimentary friction capability is included to account for the friction induced torque and friction induced
lateral loads that occur with a spinning shaft. However, friction can be neglected, in which case the solution
is consistent with that for a nonrotating shaft in a frictionless housing.
The contact stiffness can be input either as a function of time or as a function of penetration. In both cases,
the stiffness curves are input on TABLEDi Bulk Data entries which are referenced by the NLRGAP entry.
Entering stiffness as a function of time allows the user to model special situations where structural changes
over time cause the contact stiffness to vary over time. Entering stiffness as a function of penetration allows
the user to input a nonlinear stiffness curve which might be required, for example, if the shaft or housing were
covered by some nonlinear elastic material. As with stiffness, both the coefficient of friction and the clearance
can be functions of time. The time-varying coefficient of friction might be used where the quality of
lubrication varies with time, or if experimental data shows that the surface characteristics of the shaft and
housing vary with time. The clearance can be input as a function of time to model situations where an
accelerating, spinning rotor expands due to centripetal loads.
The NLRGAP entry can also approximate the torque on the shaft and housing due to friction. If this effect
is desired, the RADIUS field of the NLRGAP must be specified. If not specified, then friction effects will still
be included in the lateral loads. The torque loads will not be computed nor applied if RADIUS is input as
0.0 or left blank. As with the NOLINs, the NLRGAP is selected by the NONLINEAR Case Control
command.
NLRGAP Nonlinear Load Proportional to Gap

Defines a nonlinear radial (circular) gap for transient response or nonlinear harmonic response.

Format:
1 2 3 4 5 6 7 8 9 10
NLRGAP SID GA GB PLANE TABK TABG TABU RADIUS

Figure 4-2 NLRGAP Bulk Data Entry

Main Index
 Chapter 4: Damping 63
Nonlinear Bearings

Figure 4-3 NLRGAP Implementation

The equations underlying the NLRGAP measure the relative radial displacement between the shaft grid (Grid
A) and the housing grid (Grid B) in the displacement coordinate system. When the relative radial
displacement is greater than the clearance (i.e., when contact has occurred), contact forces as well as frictional
forces are automatically applied to the grids. The contact and frictional forces are dependent on the contact
stiffness and the coefficient of friction.
Variables used in the analysis are described below:
K(t,u) Contact stiffness, function of time or displacement.
G(t) Radial clearance, function of time.
μ(t) Coefficient of friction, function of time. Positive value implies shaft rotation from axis 1 towards
axis 2.
R Shaft radius for friction induced torque calculation.

It is helpful to visualize Grid A as belonging to the shaft and Grid B as belonging to the housing. Consider
each grid to have DOF 1 and 2 in the plane of action (e.g., DOF 1 = X and DOF 2 = Y for XY orientation).
The element relative displacement in directions 1 and 2 are:

ΔU 1 = U A1 − U B1
ΔU 2 = U A 2 − U B 2

The relative radial displacement is:

Δr = Δ U 12 + Δ U 2 2

The force in the gap when it is open ( Δ r ≤ G ( t ) ) is zero. When the gap is closed ( Δr > G (t ) ), the penetration
is defined as:

Main Index
 64 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

P = Δ r − G (t )

The nonlinear gap forces (S) are given by:

 G (t ) 
S = K (t , u )  1 − 
 Δr 

When transformed to the grid point coordinate system in direction 1 and 2, the nonlinear forces produced
by the NLRGAP element are given by:

N A1 = −ΔU 1 S + ΔU 2 S μ (t )
N B1 = − N A1 = ΔU 1 S − ΔU 2 S μ (t )
N A 2 = −Δ U 2 S − ΔU 1 S μ (t )
N B 2 = − N A 2 = Δ U 2 S + Δ U 1 S μ (t )

If the shaft radius R is input, then the following friction induced torque loads are also generated. The
expression for torque is given by (acting along 6th component for the force vector):

N A 6 = − R ΔrS μ (t )
N B 6 = − N A 6 = RΔrS μ (t )

Two Dimensional Linear/Nonlinear Element, CBUSH2D

The CBUSH2D entry can be used to define a two-dimensional connector element with an option to define
stiffness, viscous damping, and acceleration-dependent force constants in the two directions via the
PBUSH2D bulk data entry. The input format for CBUSH2D is shown in Figure 4-4.
CBUSH2D 2-D Linear-Nonlinear Connection

Defines the connectivity of a two-dimensional Linear-Nonlinear element.

Format:
1 2 3 4 5 6 7 8 9 10
CBUSH2D EID PID GA GB CID PLANE

Figure 4-4 CBUSH2D Bulk Data Card

Frequency Dependent Coefficients

It is possible to define a user-defined SCA-object to calculate properties of CBUSH2D elements in SOLs 128,
129, and 400. The user interface for the user-defined properties of a CBUSH2D consists of the associated
PBUSH2D and an ELEMUDS entry, which defines the data which is to be passed to the user routine. In

Main Index
 Chapter 4: Damping 65
Nonlinear Bearings

addition, the SCAservice must be attached to the run using the CONNECT SERVICE statement, as shown
here in the FMS (File Management) section:
CONNECT SERVICE GR1 ‘SCR.MDSolver.Obj.Uds.Elements.thpad’
The format for ELEMUDS is shown in Figure 4-5 and this entry as part of a test deck for rotordynamic
analysis is shown in Figure 4-6.

ELEMUDS Element Property User Defined Service or Subroutine

Allows the user to provide element property routines for use with specified Nastran elements.

Format:
1 2 3 4 5 6 7 8 9 10
ELEMUDS PID PTYPE GROUP UNAME DEPEN
“INT” IDATA1 IDATA2 IDATA3 IDATA4 IDATA5 IDATA6 IDATA7
IDATA8 IDATA9 ... ... IDATAn
“REAL” RDATA1 RDATA2 RDATA3 RDATA4 RDATA5 RDATA6 RDATA7

Figure 4-5 Format for ELEMUDS Entry

Figure 4-6 Test deck with ELEMUDS in Rotordynamic analysis

Main Index
 66 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

THPAD using CBUSH2D

In addition, it is possible to use the ROMAC (University of Virginia Rotating Machinery And Controls
laboratory, http://www.virginia.edu/romac/) THPAD routine (which must be obtained from ROMAC) to
calculate the properties of a tilting pad journal bearing (Branagan et al., ).
The bulk data entries required for introducing the THPAD bearing routine in analysis is shown in Figure 4-7
and Figure 4-8.

Figure 4-7 Bulk Data Cards for introducing THPAD

Figure 4-8 Usage of THPAD in rotordynamic analysis

Nonlinear Squeeze Film Damper, NLRSFD

The requirement of high power output for modern gas turbine engines has resulted in highly flexible light
weight rotor designs. Control of vibration response of these engines is a major design problem. The use of
rolling element bearings with low inherent damping make it difficult to reduce vibration amplitudes and
dynamic loads transmitted to the rotor supporting structure. Squeeze Film Dampers (SFDs) are therefore
used to provide adequate damping to maintain low amplitude vibration levels and to reduce the dynamic
loads transmitted to the bearings and rotor support structures. A general SFD model has been successfully
incorporated into the MSC Nastran time domain analysis, and this new capability provides the means to

Main Index
 Chapter 4: Damping 67
Nonlinear Bearings

design and analyze SFDs for general rotor orbits with multiple frequency content. The new capability
includes static loads and models the lift-off phenomenon important in the design of free-floating dampers.

NLRSFD Nonlinear Transient Load Proportional to a Squeeze Film Damper

Defines a nonlinear transient radial squeeze film damper.

Format:
1 2 3 4 5 6 7 8 9 10
NLRSFD SID GA GB PLANE BDIA BLEN BCLR SOLN
VISCO PVAPCO NPORT PRES1 THETA1 PRES2 THETA2 NPNT
OFFSET1 OFFSET2 GROUP NAME2
NAME
RDATA1 RDATA2 RDATA3 RDATA4 RDATA5 RDATA6 RDATA7 RDATA8

Figure 4-9 Bulk data entry for NLRSFD

As described earlier, the general SFD model incorporated in MSC Nastran for transient analysis inserts the
SFD forces in the right hand side of the equations as a nonlinear force, {N(t)}. The bulk data entry for SFD
is called NLRSFD and is shown in Figure 4-9. Input data required for SFD definition are: journal diameter
and clearance, land length, oil viscosity, and location of ports and boundary pressure at these points.

Equations for SFD

A representative model for the squeeze film damper is shown in Figure 4-10. Here Grid I represents a point on
the damper journal (rotor) and Grid J is a point on the housing (stator). The two grid points should be
coincident and have parallel Cartesian coordinate systems. The SFD code uses relative displacements and
velocities, x , y , x , y between these two grid points as the input and outputs the forces FX ( x , y , x , y ) and
FY ( x , y , x , y ) acting on the damper journal grid point (Grid I). Equal and opposite forces: − F X ( x , y , x , y )
and − FY ( x , y , x , y ) are applied to the SFD housing grid point (Grid J). If parallel centering is required, then
this can be modeled using the CELAS2 or CBUSH elements.

Main Index
 68 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

Figure 4-10 SFD Model in MSC Nastran

The squeeze-film damper model is based on work originally performed at Case Western Reserve University
(CWRU). It incorporates a numerical solution of the Reynolds lubrication equation for incompressible
laminar isoviscous films that is described in Adams, Padovan and Fertis (Adams et al., 1982). The model is
capable of handling the specified pressure boundaries at the feed (supply) and discharge (drain) ports of the
SFD. The SFD pressure distribution is determined using a one-dimensional finite difference scheme. The
one-dimensional finite difference approach accounts for static as well as dynamic deflections and is capable
of modeling general damper orbits with a broad frequency content. The model computes the oil film forces
by numerical integration of the instantaneous film pressure distribution.
SFD Model in MSC Nastran also shows the SFD model coordinate system and the supply and discharge port
locations. Either one or two ports can be used to define pressure boundary conditions for the SFD model.
The number of grid points used in the finite difference scheme to determine the film pressure distribution is
defined by the variable NGRID. For the single port model, there are NGRID points, and for the double port
model there are NGRID points between the two ports for a total of 2 x NGRID points. Typically, a value of
about 31 (an odd number is required) for NGRID will be adequate.

Main Index
 Chapter 4: Damping 69
Nonlinear Bearings

Figure 4-11 SFD in operation

A SFD in operation is shown in Figure 4-11, where e is the whirl vector which shows displacement of Grid I
on the rotor from Grid J on the center of housing. Using this vector, displacement and velocities between the
two points can be written as:

x =| e | cos(Ωt )
y =| e | sin(Ωt )
x = −Ω | e | sin(Ωt )
y = Ω | e | cos(Ωt )

These values are provided as input to the SFD code for the rest of the analysis.

Squeeze Film Damper in CBUSH2D

For better accuracy and to facilitate use in other solution sequences the NLRSFD was also implemented as
an element. The Squeeze Film Damper was added as an option of a more general 2-D bearing element
CBUSH2D.
The format for the PBUSH2D entry with added options to define a SFD is shown in Figure 4-12.

Main Index
 70 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

PBUSH2D 2-D Linear/Nonlinear Connection Properties

Defines linear and nonlinear properties of a two-dimensional element (CBUSH2D entry).

Format:
1 2 3 4 5 6 7 8 9 10
PBUSH2D PID K11 K22 B11 B22 M11 M22
SQUEEZE BDIA BLEN BCLR SOLN VISCO PVAPCO
NPORT PRES1 THETA1 PRES2 THETA2 OFFSET1 OFFSET2

Figure 4-12 PBUSH2D entry with option for SFD

Nonlinear Load Functions (NOLINi)

The nonlinear load functions (NOLIN1, NOLIN2, NOLIN3 and NOLIN4) are available in all of the
transient solutions for the purpose of generating direct forces from simple displacement and velocity
functions. This is a mature capability in MSC Nastran that preceded the development of the material and
geometric nonlinear capabilities. It was intended to provide a direct method for modeling certain
mechanisms and special effects at the local level where only a few degrees of freedom are coupled. It is not
intended for general nonlinear analysis since the logic cannot detect geometric changes and has no means of
using element stresses or forces. The main applications for the NOLINi functions are for local devices such
as joints and nonlinear dampers. For each nonlinear force component, the dependencies with displacement
and velocity degrees-of-freedom are explicitly defined on one or more bulk data entries.
The limitations on the use of these functions are caused primarily by their simplicity. These limitations are:
1. They have no memory or other path dependencies. Specifically, they provide no direct means for
including data from previous states.
2. Each input function only applies loads to a single degree of freedom, which can require a great deal
of effort to describe complex models.
3. The system treats these functions as simple forces rather than finite elements. Therefore, the lack of
a tangent matrix results in potential stability problems.
4. In the linear solutions, all degrees of freedom in the NOLIN functions must be in the dynamic
solution set.
5. In the linear solutions, the forces created by NOLIN entries lag by one time step. A common way to
attempt to account for this is to decrease the integration time step.

The input data starts with a case control request: NONLINEAR (Case), with one or more NOLINi Bulk
Data entries to define the set. The four basic options to define a scalar nonlinear force, Ni, in terms of the
degrees of freedom, {gj}, are summarized in the table below.

Main Index
 Chapter 4: Damping 71
Nonlinear Bearings

Table 4-1 Options for Nonlinear Load using NOLIN

Table 4-2
Option Function Comments
NOLIN1 Ni = SF(gj) F(gj) is a TABLEDi input
NOLIN2 Ni = Sgjgk Product of two variables
A
NOLIN3 Ni = S(gj) , gj > 0 A is an input exponenet
A
NOLIN4 Ni = -S(-gj) , gj < 0 As 3 except for negative g

The variables, gj, may be displacement or velocity components of grid, scalar, or extra points in the solution
set. In the DIRECT TRANSIENT solutions, the connected degrees of freedom, gi and gj, etc., must remain
in the solution set. In the modal transient solutions, only extra points are available for use by the NOLINi
entries. In nonlinear transient solutions, all degrees of freedom are available; however, extra points are not
supported. Note that these functions mimic the basic nonlinear function generators used in passive analog
computers. They may be added together and combined with other functions such as MPCs and scalar
elements to handle a variety of problems.
The bulk data format for the NOLIN1 entry is shown in Figure 4-13 . Other NOLIN options follow a similar
format.

NOLIN1 Nonlinear Load as a Tabular Function

Defines a forcing function for transient response or nonlinear harmonic response of the form

Format:
1 2 3 4 5 6 7 8 9 10
NOLIN1 SID GI CI S GJ CJ TID

Figure 4-13 NOLIN1 Bulk Data Entry

Main Index
 72 MSC Nastran Rotordynamics User’s Guide
Nonlinear Bearings

Main Index
 Chapter 5: Rotor Usage
MSC Nastran Implicit Nonlinear (SOL 600) User’s GuideRotordynamics User’s Guide

5 Rotor Usage


Rotor Usage 74

Case Control Entries 74
 Parameters 74

Bulk Data Entries 75

Main Index
74 MSC Nastran Rotordynamics User’s Guide
Rotor Usage

Rotor Usage
Rotor modeling techniques available in MSC Nastran were discussed in Chapter 3. In order to perform
complex eigenvalue, frequency response, or transient analysis, some other entries need to be defined in Case
Control and Bulk Data. Most of these entries are common to the three types of rotors discussed in the
previous chapter. Brief overview of these entries is provided in rest of the chapter. More details are available
in the MSC Nastran Quick Reference Guide.

Case Control Entries

RGYRO
The RGYRO Case Control command activates the rotordynamics capability, and selects the RGYRO Bulk
Data entry for use in complex modes, frequency response, and static analysis. For transient response, the
RGYRO command selects the UNBALNC (described later in this chapter) Bulk Data entry. If the
UNBALNC entry is not required, setting RGYRO to YES will include the gyroscopic effects in the transient
response calculation. Setting RGYRO to NO will deactivate gyroscopic effects in all solutions.For more
information see, RGYRO (Case).

CAMPBELL
This entry specifies parameters for the generation of the Campbell Diagram in complex eigenvalue analysis
(SOL 107). When there is a Case Control request for the Campbell diagram, the selected RGYRO Bulk Data
entry (described later in this chapter) must have the ASYNC option specified in its SYNCFLG field (Field
3). Otherwise, the program terminates the execution with an appropriate fatal message. For more information
see, CAMPBELL (Case). In V2018, second row was introduced in CAMPBLL entry to enable eigen-vector based
mode tracking.

Parameters

WR3, WR4, and WRH

The nature of damping to be used for rotordynamic analysis can be controlled based on the definition of
following parameters for complex eigenvalue analysis and transient response analysis. For frequency response,
it is dependent on these values and the parameter GYROAVG. For more information see, WR3, WR4, WRH.

GYROAVG
The parameter GYROAVG is used to specify one of the two formulations for frequency response analysis
using the rotordynamic capability. The default is to determine any frequency-dependent terms for each
frequency. This option activates the frequency-dependent looping option. Setting the value < 0 uses an
‘average’ frequency formulation. This option avoids using the frequency-dependent looping and results in a
 Chapter 5: Rotor Usage 75
Bulk Data Entries

shorter execution time. For this option, PARAM WR3 /WR4 /WRH must be specified to include rotor
damping. For more information see, GYROAVG.

WHIRLOPT
When the analysis is performed in rotating reference frame, (only available with ROTOR entry), the
synchronous analysis for forward and backward whirl modes follow different set of equations of motion. The
user has the option to select required whirl mode option using this parameter. Here, the forward whirl analysis
is treated as the default option. For more information see, WHIRLOPT.

Bulk Data Entries

The RGYRO and RSPINR/RSPINT entries are honored and processed only in residual or assembly jobs and
are ignored in external superelement creation jobs.

RGYRO
The RGYRO entry in Bulk Data includes identification number that is selected by the Case Control
command. Only one RGYRO entry can be selected at a time, and the others specified in the input file are
ignored. This card also specifies whether the analysis is synchronous or asynchronous analysis using the
“SYNCFLG” field. This information is a required input for frequency response and complex modes analyses.
In case of multiple rotors present in the analysis, the “REFROTR” field specifies the reference rotor ID for
the analysis. The “SPUNIT” field specifies whether the entries SPDLOW, SPDHIGH, and SPEED are given
in terms of RPM (revolutions/minute) or frequency (revolutions (cycles)/unit time). When there is a Case
Control request for a Campbell diagram, the selected RGYRO Bulk Data entry must have the ASYNC option
specified in its SYNCFLG field.
The second row defines whirl frequency for calculation of rotor damping and circulation terms for rotor
structural damping specified through GR field for asynchronous complex eigenvalue analysis only. In case any
of the whirl frequency parameters (WR3WRL) are defined along with corresponding damping parameter
(WR3 or WR3R), an appropriate FATAL message is produced. For more information see, RGYRO.
RGYRO Rotordynamic Analysis Parameters

Specifies synchronous or asynchronous analysis, reference rotor, and rotation speed of the reference rotor.

Format:
1 2 3 4 5 6 7 8 9 10
RGYRO RID SYNCFLG REFROTR SPDUNIT SPDLOW SPDHIGH SPEED ROTRSEID

WR3WRL WR4WRL WRHWRL

Main Index
76 MSC Nastran Rotordynamics User’s Guide
Bulk Data Entries

RSPINR
The RSPINR entry specifies the relative spin rates between rotors and identifies the axis of rotation for
complex eigenvalue, frequency response, and static analysis. The positive spin direction is from Grid A to
Grid B. For each rotor defined in the model (either in residual or in external superelement), a corresponding
RSPINR entry is required. For rotor defined in external SE, the GRIDA and GRIDB specified in RSPINR
must be part of the SEs ASET during creation run. This requirement is relaxed for rotors defined in PART
SE. If SPTID is a real number, the value is considered constant. The SPTID field specifies relative spin rate.
If SPTID is real, it is used directly as scaling factor for rotor speed. If SPTID is an integer number, the value
references a DDVAL entry that specifies the relative rotor spin rates. The number of spin rates for each rotor
must be the same. Relative spin rates are determined by correlation of table entries. The ith entry for each
rotor specifies the relative spin rates between rotors at RPMi/FREQi. Spin rates for the reference rotor must
be in ascending or descending order. Based on the value provided for SPTID, the expression for rotor speed
in terms of reference rotor speed is given by:
Ω = α + βΩref
where α and β are determined by a least-mean-square fit of the relative rotor speeds input on the RSPINR
between SPDLOW and SPDHIGH of the reference rotor (specified using RGYRO entry), and Ωref is the
rotor speed for reference rotor.
The RSPINR entry is also used to specify damping options and structural damping parameters for the rotor
(except Hybrid damping). The WR3R, WR4R and WRHR parameters specify “average” excitation
frequency for calculation of rotor damping and circulation terms for rotor structural damping specified
through GR field, through material entries and through hybrid damping card ROTHYBD, respectively. The
rotor damping is included in the analysis as described in Chapter 4. For more information see, RSPINR.

RSPINR Relative Spin Rates Between Rotors

Specifies the relative spin rates between rotors for complex eigenvalue, frequency response, and static
analysis.

Format:
1 2 3 4 5 6 7 8 9 10
RSPINR ROTORID GRIDA GRIDB SPDUNT SPTID ROTRSEID

GR ALPHAR1 ALPHAR2 WR3R WR4R WRHR

RSPINT
The RSPINT entry performs similar function as RSPINR for transient analysis. For transient analysis, the
SPTID field references a TABLED1 entry that specifies the rotor spin rate history. For more information see,
RSPINT.
 Chapter 5: Rotor Usage 77
Bulk Data Entries

RSPINT Rotor Spin Rates

Specifies rotor spin rates for transient analysis.

Format:
1 2 3 4 5 6 7 8 9 10
RSPINT ROTORID GRIDA GRIDB SPDUNT SPTID SPDOUT ROTSEID
GR ALPHAR1 ALPHAR2 WR3R WR4R WRHR

ROTHYBD
This entry is used to introduce hybrid damping option for rotors. The ROTORIDi – HYBDAMPi pair
values referencing nonexistent rotors are ignored. Hybrid damping can result in very densely populated
damping matrix causing significant performance penalty. Details about the matrices generated hybrid
damping is given in HYBDAMP card in QRG. For more information see, ROTHYBD.

ROTHYBD Hybrid damping for rotors

Specifies hybrid damping data for rotors.

Format:
1 2 3 4 5 6 7 8 9 10
ROTHYBD ROTORID1 HYBDAMP1 ROTORID2 HYBDAMP2

CAMPBLL
This entry is required for the generation of a Campbell Diagram using complex eigenvalue analysis. The
CAMPBLL option is supported in both SOL 107 and SOL 110. However, it is not supported in SOL 200
and SOL 400 for ANALYSIS=DCEIG and ANALYSIS=MCEIG. The rotor speed at which the complex
eigenvalue analysis needs to be performed for a Campbell Diagram are specified using the DDVALID field
and appropriate speed units in the TYPE field.
In V2018, two different approaches were implemented for mode tracking while generating Campbell
diagram. In both the methods, only the complex frequencies with a positive imaginary part are considered
for mode tracking.

Main Index
78 MSC Nastran Rotordynamics User’s Guide
Bulk Data Entries

Numerical Approach
This method is used for mode tracking once the eigenvalues have been extracted for all the rotor speeds
defined for analysis using CAMPBLL and DDVAL bulk data entries. For each rotor speed, the second
derivative is computed using the central difference approach.

Eigenvector-Based Approach
Unlike the numerical approach, here both left and right eigenvectors associated with complex eigenvalues are
used for mode tracking. The method implemented is described in detail in Reference 2. As described in the
paper, Normalized Cross Complex Orthogonality (NC2O) parameter is computed for each pair frequencies
for two rotor speeds. In V2018, the eigenvector based mode tracking approach is available for analysis in
FIXED reference frame only when HESS option is used in EIGC for eigenvalue extraction. Due to
limitations of the HESS option when working with large size models without a reduction, the use of SOL
110 or a superelement based reduction approach is recommended for analyzing large rotor models.
For more information see, CAMPBLL.

CAMPBLL Campbell Diagram Parameters

Specifies the parameters for Campbell diagram generation.

Formats:
1 2 3 4 5 6 7 8 9 10
CAMPBLL CID VPARM DDVALID TYPE
MODTRK CORU SWTTR NUMMOD PRTCOR

CAMPBLL 15 SPEED 22 RPM

1 0.75 1
 Chapter 5: Rotor Usage 79
Bulk Data Entries

DDVAL
This entry is required for generating a Campbell diagram, and it defines real rotor speed for asynchronous
complex eigenvalue analysis. For more information see, DDVAL.
DDVAL Discrete Design Variable Values

Defines real, discrete design variable values for use in discrete variable optimization, topometry optimization
or rotordynamics.

Format:
1 2 3 4 5 6 7 8 9 10
DDVAL ID DVAL1 DVAL2 DVAL3 DVAL4 DVAL5 DVAL6 DVAL7

UNBALNC
The UNBALNC entry is used to define rotating unbalanced loads for rotordynamic transient and frequency
response analysis. For transient response, UNBALNC is specified by the RGYRO Case Control command.
For frequency response, it is specified by the DLOAD Case Control command. The imbalance load will be
generated based on the mass value (Field “MASS”), offset values (Field “ROFFSET” and “ZOFFSET”), and
the rotor spin speed specified using RGYRO and RSPINR/RSPINT.
In the case of ROFFSET field, if the entry is a real number, the value is considered constant. If the entry is
an integer number, the value references a TABLEDi entry that specifies the value as a function of time for
transient response or frequency for frequency response. The Theta is measured from the plane defined by the
z-axis and the user specified vector (X1, X2, X3). Theta = 0.0 is in the direction of the user-specified vector
(X1, X2, X3). Unbalance rotation is in the positive z-direction.
For frequency response analysis, the EPOINTs and the continuation cards are ignored. For transient analysis
in fixed reference frame, CFLAG field can be used include mass correction or speed correction or both
corrections in the analysis. The EPOINTS specified in the fourth and fifth row on the UNBALNC entry can
be used to extract these correction forces and moments. For more information see, UNBALNC.

Main Index
80 MSC Nastran Rotordynamics User’s Guide
Bulk Data Entries

UNBALNC Specifies an Unbalanced Load for Rotordynamic Transient or Frequency Response

Analysis

Used in rotordynamic analyses to specify a rotating unbalance load in terms of a cylindrical system with the
rotor rotation as the z-axis.

Format:
1 2 3 4 5 6 7 8 9 10
UNBALNC SID MASS GRID X1 X2 X3
ROFFSET THETA ZOFFSET T/Fon T/FOFF CFLAG
UFT1 UFT2 UFT3 UFR1 UFR2 UFR3
MCT1 MCT2 MCT3 MCR1 MCR2 MCR3
SCR1 SCR2 SCR3

Forces due to imbalance at grid “k” in rotor coordinate system for transient analysis are given by (where the
rotor axis is along Z axis):

 Fx (t )   m (t ) r (t ) ( Ω 2j (t ) cos θ jk (t ) + Ω
 (t ) sin θ (t ) ) 
 j jk 
  = WTMASS .  
F ( t )  m (t ) r (t ) ( Ω j (t ) sin θ jk (t ) − Ω j (t ) cos θ jk (t ) ) 

2
 y 

where
m(t) Specified by the user (MASS field in UNBALNC)
r(t) Specified by the user (ROFFSET field in UNBALNC)
Ωj ( t )
Specified by the user through RSINT/RSPINR entry
 (t )
Ω j Derived from Ω j (t )
θ jk ( t )
Instantaneous angular location, measured from axis 1 to axis 2 (=  Ω j (t ) dt + θ k )
0
θk
Specified by the user (THETA field in UNBALNC)
j Rotor on which the grid is located

In the rotating coordinate system, similar equation for unbalance loads is used.
 Fx (t )   m(t )r (t )Ω j (t ) cos θ k 
2

  = WTMASS .  
 m(t )r (t )Ω j (t ) sin θ k 
2
 Fy (t ) rot
 Chapter 5: Rotor Usage 81
Bulk Data Entries

Correction terms for transient analysis in fixed reference frame

When the CFLAG field is prescribed in the UNBALNC entry, additional correction loads are also generated
in addition to the standard unbalance loads. These correction loads include contribution from unbalance
mass to the inertia and gyroscopic terms in the analysis.

 Fx (t )   Fx (t )   − mk (t )ux (t ) 
     
 Fy (t )  =  Fy (t )  + WTMASS .  − mk (t )uy (t ) 
 F (t )     − m (t )u (t ) 
 z Corr 0   k z 

where displacements are transformed from the global to rotor-force coordinate. The above correction
accounts for the change in the translation mass due to the imbalance mass (CFLAG = ‘MASS’).
Similarly, the correction to moments is given by:

M x (t )  M x (t )   −mk (t )(rk (t )sin θ jk (t ))2 (φx (t ) + 2Ω jφx (t ) + Ω

 φ (t )) 
j x
     2 

 
M y (t ) = M y (t ) + WTMASS.  −mk (t )(rk (t )cosθ jk (t )) (φy (t ) − 2Ω jφy (t ) − Ω jφy (t )) 
M (t )      
 z Corr 0   −mk (t )rk (t )φz (t )
2


where φ is the finite rotation about the i-axis.

In the above equation, the first correction term accounts for the change in rotational inertia due to the
imbalance mass, the second term corrects the gyroscopic matrix, BG. These terms are added if CFLAG on the
UNBALNC entry is ‘MASS’ or ‘BOTH’. The third term corrects the tangent matrix, KT. This term is added if
CFLAG is ‘SPEED’ or ‘BOTH.
The forces and moments are transformed from the rotor-force coordinate system using the transformation
matrix stored on the UNBALNC entry.
For frequency response analysis, the unbalance load in fixed reference frame given by:

 Fx (ω )   mrω ( cos θ k + i sin θ k ) 

2

  = WTMASS .  
 Fy (ω )   mrω ( sin θ k − i cos θ k ) 
2

For frequency response analysis, the unbalance load in rotating reference frame given by:

 Fx (ω )   mrω ( cos θ k ) 
2

  = WTMASS .  
 Fy (ω )   mrω ( sin θ k ) 
2

Main Index
82 MSC Nastran Rotordynamics User’s Guide
Bulk Data Entries
 Chapter 6: Rotordynamics with MSC Patran
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Rotordynamics with MSC

6 Patran


Introduction 84
 Example Model 85

Main Index
 84 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
For the first time for MSC Nastran Rotordynamics users, use of MSC Patran for developing the basic model
is encouraged. Once the baseline model is created using MSC Patran, other rotordynamics related features
can be easily manually included in the input file.

MSC Patran Set up

If you are new to Patran, here is a little configuration trick that is particular to Windows OS and that may
help you organize many more models that you will build now and in the future. Find the Patran icon on your
system. (To build this document, we worked with Patran 2014.1 64-bit.)

Right-click this icon and drag & drop it into your desired working directory and left-click Create shortcut
here. For this example, we chose > Libraries > Documents > Patran > RotorA on our system.

Figure 6-1 Customizing Patran Setup

Main Index
 Chapter 6: Rotordynamics with MSC Patran 85
Example Model

Now, right-click this newly created shortcut to Patran and left-click Properties. Delete the path in the Start
in: field which is the one that was chosen during installation on your system. So the Start in: field should
now be blank as shown. Click OK.
So, every time you start Patran from this particular shortcut, the default directory will be RotorA. While one
could alternatively manage Patran’s default directory, this shortcut technique is found to be very useful.

Example Model
As a first example, a hythothetical hollow cantilever rotor model is developed using Patran. The hollow rotor
is 0.5m long, and has an inner diameter of 0.05m and outer diameter of 0.1m, as shown in Figure 6-2. The
rotor is made up of steel and is spinning at 10,000 RPM. In this example, the rotor needs to be modeled
using 5 CBEAM elements and Node 1 is constrained.

Figure 6-2 Example hollow rotor model

Main Index
 86 MSC Nastran Rotordynamics User’s Guide
Example Model

Patran Model
Here are the main steps involved in the development of the Patran model. These steps are described in more
details in the later sections. Steps 1 to 6 are standard for any MSC Nastran model and further details about
these steps are available in the MSC Nastran’s Linear Statics Users Guide and in the MSC Nastran’s Dynamics
User`s Guide. Step 7 and Step 8 are required only if the model contains rotors. Currently, MSC Patran
supports 1D rotors defined using ROTORG only.

Step 1 Initializing Patran

Step 2 Developing Geometry in Patran
Step 3 Defining material properties in Patran
Step 4 Defining 1D Beam Properties
Step 5 Generating FE mesh
Step 6 Defining constraints for the model
Step 7 Introducing rotor definition
Step 8 Setting up Analysis
Step 9 Running MSC Nastran

For Post-Processing, MSC Patran supports generation of mode shapes and animations for the eigenvectors in
complex eigenvalue analysis. If the input file includes an option for generating Campbell Diagram, then it
can also be plotted. These features are described in Steps 10 and 11.

Step 10 Generating Mode Shapes

Step 11 Generating Campbell Diagram

Step 1: Initializing Patran

1. Double-click the Patran shortcut in the RotorA directory that we just created.
2. Select File menu -> New and specify File name: to be rotor. Click OK.
Before any actions can be carried out, Patran requires the definition of a database. Notice how the directory
defaults to RotorA now, see Figure 6-3.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 87
Example Model

Figure 6-3 Defining Patran File

3. In the New Model Preference tab, select Tolerance -> Default, as shown in Figure 6-4.
4. Click OK

Main Index
 88 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-4 Setting up model tolerance

Step 2: Developing Geometry in Patran

1. Click the Geometry tab
2. Select Curves -> select Line By XYZ (see Figure 6-5(a))
The previous two steps populate the Geometry tool with the following choices: Action: Create, Object:
Curve, and Method: XYZ as shown in Figure 6-5(b). Alternatively, the same Action-Object-Method sequence
can be obtained by using the individual dropdown menus within the Geometry tool on the right.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 89
Example Model

Figure 6-5 Creating geometry

3. Specify the length of the rotor which measures 0.5 meters (see Figure 6-5(b)). Accepting MSC
Nastran’s basic coordinate system, specify a line from the origin to the <X Y Z> coordinate <0.5 0 0>.
(see Figure 6-5(b))
4. Click Apply.
Patran now shows this line in the GUI, as shown in Figure 6-6.

Figure 6-6 Line Geometry

Main Index
 90 MSC Nastran Rotordynamics User’s Guide
Example Model

Step 3: Defining material properties in Patran

1. Begin defining the rotor’s material and section properties by clicking the Properties tab and the
Isotropic button in succession, see Figure 6-7.

Figure 6-7 Main menu to define material properties

This action invokes the materials tool, as shown in Figure 6-8.

2. Specify the Material Name as ‘steel’ as shown in Figure 6-8.
3. Click the Input Properties button and specify Elastic Modulus (210.e9), Poisson’s Ratio (.3) and
Density (7810.) as shown in Figure 6-8.
4. Click OK to close the Input Options window.
5. Click Apply in the Materials tool.
Patran relies heavily on the Apply button and you want to be certain to click it here and every time you define
something new in your model. If you are working with a limited screen resolution, you might need to scroll
down to find the Apply button. Check that the material steel should now be listed among your listing of
Existing Materials as shown in Figure 6-9.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 91
Example Model

Figure 6-8 Entering Material Properties

Figure 6-9 Material property definition

Main Index
 92 MSC Nastran Rotordynamics User’s Guide
Example Model

Step 4: Defining 1D Beam Properties

1. Click on the Properties tab and then on the Beam tool within the 1D Properties tool group as shown
in Figure 6-10.

Figure 6-10 Main menu to define 1D beam properties

2. Specify the beam_prop in the Property Set Name field, see Figure 6-11.
3. Select the General Section (CBEAM) element option in the Options: dropdown.
4. Click the Input Properties button in the tool interface, place the cursor in the Material Name field
and click the Mat Prop Name icon to open the Select Material interface. Select the previously defined
‘steel’ entry and verify that the Material Name field now contains the ‘m:steel’ string.
5. The Bar Orientation indicates the bending direction of the beam. Since our beam is directed along
the x-axis, we can either specify Y (i.e. vector <0,1.,0>), as shown, or Z (i.e. vector <0,0,1.> . Here, Y
direction is selected as the bending direction by specifying vector <0,1,0> in the Bar Orientation
field.
6. Since the rotor is represented by a hollow tube, we use the built-in beam library for convenience.
Access the Beam Library by clicking its icon. Beam Library that opens is shown in Figure 6-12.
7. The “less than” and “greater than” icons allow you to scroll through the library. The first click on the
“greater than” icon reveals the appropriate icon for the parametric representation of a hollow tube in
the top left corner of the displayed 3x3 icon matrix.
8. Click on the hollow cross section icon and specify the New Section Name as “rotor_section”.
9. Enter the values for the respective outer and inner radii R1 = 0.1 and R2 = 0.05.
10. Click the Apply button to save this particular section among the Existing Sections in this model.
11. Click the Calculate/Display button to verify the beam section. This presents cross sectional details,
as shown in Figure 6-13.
12. Close the Section Display window.
13. Click OK to close the Beam Library window.
Notice that rotor_section now populates the [Section Name] field, as shown in Figure 6-14.
14. Click OK to close the Input Properties window.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 93
Example Model

Figure 6-11 Defining beam properties

Main Index
 94 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-12 Options in Patran to select beam cross section

Figure 6-13 Cross section for the hollow beam

Main Index
 Chapter 6: Rotordynamics with MSC Patran 95
Example Model

Figure 6-14 Final Beam properties

Step 5: Generating FE mesh

1. Click the Meshing tab to start the meshing process. In order to obtain a uniform mesh and control
the meshing process, mesh seeds will be generated first and then beam elements will be generated. (see
Figure 6-15)
2. Select Action: Create, Object: Mesh Seed, Type: Uniform. Then enter “5” as the number of mesh
seeds to be defined along the curve since we are planning to generate 5 beam elements. And finally
define the curve where the mesh seeds need to be generated (see Figure 6-16). Now, we will generate
the actual mesh for the curve.

Figure 6-15 Main Menu to generate finite element mesh

Main Index
 96 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-16 Generation of mesh seeds along a curve

3. Select Action: Create, Object: Mesh, Type: Curve. (see Figure 6-17)
4. MSC Patran allows the user to control the starting node ID and elements ID while meshing. The
default value used is 1.
5. Set the curser in the Curve List input field and click on the line in the GUI. This populates the field
with the ID ‘Curve 1’ as shown. (Alternatively, you could simply type ‘Curve 1’ into that field.)
6. Uncheck Automatic Calculation in the Global Edge Length category and set the value to 0.1 as
shown. (Note that since we have already generated mesh seeds for the current curve, this value has no
significance. However, it can be used to control mesh size for more complicated geometries.)
7. Click the Select Existing Prop button and choose the earlier defined beam_prop set. This step
associates the entire beam_prop set with the CBEAM card and its underlying element formulation.
Click Apply.
8. Click on the Home tab (see Figure 6-18(a)), click the Label Control tool button. This allows us to
verify placement of nodes and elements of the previous meshing operation, as shown in Figure 6-18(b).

Main Index
 Chapter 6: Rotordynamics with MSC Patran 97
Example Model

Figure 6-17 Steps to generate beam elements

Figure 6-18 Model with beam elements

Main Index
 98 MSC Nastran Rotordynamics User’s Guide
Example Model

Step 6: Defining constraints for the model

To constrain the rotor in space, cantilever boundary conditions are defined on the left end.
1. Select the Loads/BCs tab, and then select the following to define displacement constraints:
Action: Create, Object: Displacement, Type: Nodal (Figure 6-19)
2. Choose an expedient New Set Name such as ‘cantilever_Rotor’
3. Click the Input Data specify zero translations <0,0,0> for T1, T2, and T3 (i.e. in X, Y, and Z) as well
as zero rotations <0,0,0> for R1, R2, and R3 (i.e. around X, Y, and Z).
4. Click OK.
5. Click the Select Application Region
6. In the pop-up window, select FEM as shown in Figure 6-20(a).
7. Then pick Node 1 under “Select Nodes”.
8. Click on “Add”. This populates the “Application Region” with Node 1. And click OK (See
Figure 6-20(b)).

Figure 6-19 Defining constraints for the beam model

Main Index
 Chapter 6: Rotordynamics with MSC Patran 99
Example Model

Figure 6-20 Applying constraints to the beam model

9. At the end, you should see the “Cantilever_Rotor” listed as the boundary condition under Existing
sets as shown in Figure 6-21.
10. Click Apply. The Patran display should show the boundary condition being applied to Node 1. This
verifies that all translations and rotations of this particular vertex are constrained as shown in
Figure 6-22.

Main Index
 100 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-21 Confirmation for definition of constraints

Figure 6-22 Beam model with constraints applied

Step 7: Introducing Rotor Definition

In order for MSC Patran to capture the Rotordynamic effect, we need define the constituents of the rotor
and direction of rotation.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 101
Example Model

1. Click on Tools menu in Patran’s top tool bar, select the Modeling option, and click on
Rotordynamics as shown in Figure 6-23.
2. This sets Action: Create, Analysis: Steady State, and Object: Rotor as shown.
3. Choose a Rotor Name for this rotor data set such as ‘hollow_rotor’.
4. Click the cursor into the Rotor Node List field and rubber-band select all rotor nodes. Alternatively,
you could simply type ‘Node 1:6’. You will later see that this populates the ROTORG card.
5. To specify the orientation of the rotor, select any two line rotor nodes. Node 2 and Node 3 are
specified for Spin Direction Node 2 and Spin Direction Node 3, respectively.
6. Click the Spin Profile button and specify ‘1.’ in the Speed Values field. This value is a multiplier
which we are setting to unity. Although Patran has some logic to distinguish between intended scalar
indices and decimals, it is good practice to distinguish between the two explicitly by using the decimal
point for the latter.
7. Click OK to close the Spin Profile tool tab.
8. Click Apply to commit the hollow_rotor to the Existing Rotors.
9. Click the Close button to close the Rotor Dynamics tool tab.

Figure 6-23 Selecting Rotor Dynamics tool

Main Index
 102 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-24 Steps involved in defining rotor

Step 8: Setting up Analysis

1. Click on the Analysis tab as shown in Figure 6-25. This invokes the Analysis tool in which the default
values of Action: Analyze, Object: Entire Model, Method: Full Run are appropriate.

Figure 6-25 Patran menu to define analysis

Main Index
 Chapter 6: Rotordynamics with MSC Patran 103
Example Model

Figure 6-26 (a) Setup for "Solution Type"

Main Index
 104 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-27 (b) Setup for "Solution Type"

2. As shown in Figure 6-26, enter Job Name as “hollow_rotor_test”. This will be the name of the input
deck created by Patran. You can add some descriptive SUBTITLE (e.g. CBEAM Rotor) and LABEL
(e.g. Cantilevered, Left) if you like. However, this is not absolutely necessary for the configuration of
this model.
3. Click the Solution Type.
4. Select COMPLEX EIGENVALUE and choose Direct in the Formulation drop-down menu. Verify
that Solution Sequence 107 is now posted near the bottom of the Solution Type tab. Next click on
Solution Parameters
5. Select Mass Calculations as Coupled for better accuracy. Next click on COMPLEX EIGENVALUE
tab. This opens the Eigenvalue Extraction window.
6. Here, select “Complex Lanczos” as the Extraction Method. Set the Number of Desired Roots = field
to 12. Click OK in the Eigenvalue Extraction pop-up window to close it.
7. Click on Results Output Format to control the format for output. As shown in the Results Output
Format pop-up window, the default options are XDB and Print. Click OK to close this pop-up.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 105
Example Model

8. Click OK in the Solution Parameters pop-up window to close it. Click OK on the Solution Type
tool tab to close it.

Figure 6-28 Defining rotor in analysis

Main Index
 106 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-29 Defining rotor in analysis

9. Next, click the Subcases as shown in Figure 6-28. This opens the Subcases pop-up window.
10. Since the Subcase Name is already set to ‘Default’ and the ‘Default’ in the Available Load Cases
contains all applied boundary conditions, simply click the Subcase Parameters button to modify the
Default Subcase.
11. Check Enable Rotor Dynamics in the Subcase Parameters window. Click the Specify Spinning
Properties.
12. Choose the Option: Asynchronous, select the previously defined “hollow_rotor”, set the Speed Unit
to RPM and specify Speed = 10000. This is the second speed entry that is used in RGYRO.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 107
Example Model

13. Click OK to close the Spinning Properties window. Click OK to close the Subcase Parameters
window. Click Apply to commit these changes to the Default subcase.
14. Click Cancel to close the Subcases window.

Step 9: Running MSC Nastran using Patran

1. Click on Subcases Select and ensure that Default is in the Subcases Select box as shown in Figure 6-30.
2. Click on OK in Subcases Select and click the Apply button in the Analysis tool tab to invoke the
analysis and Patran’s call to MSC Nastran.

Figure 6-30 Selection of final subcases

Some of the benefits of having forced Patran to write everything into the RotorA directory are now becoming
evident: All of the pertinent files are nicely collected here. In addition to the hollow_rotor_test.db database
file that we needed to define initially, Patran also kept a session file (i.e. *.ses.*) that tracks all of our moves,
and most importantly, Patran wrote the *.bdf input deck that is passed to MSC Nastran upon invocation.
Once the computation completes, output databases with *.MASTER, *.DBALL, and *.xdb extension are
observed. Also, it is encouraged that the user should look into the .F06 and search for the presence of FATAL
MESSAGE before further post processing.

Main Index
 108 MSC Nastran Rotordynamics User’s Guide
Example Model

Next are the steps involved in Post Processing

Step 10: Generating Modeshapes

1. Click the Analysis tab and select the XDB option in the Access Results tool group as shown in
Figure 6-31.

Figure 6-31 Menu to select results for post processing

Figure 6-32 Attaching XDB file for postprocessing

2. This invokes the Action: Access Results, Object: Attach XDB, and Method: Result Entities as shown
in Figure 6-32.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 109
Example Model

3. Next click the Select Results File and select the *.XDB file just created from the MSC Nastran run.
4. Click the OK button in Select File and Apply button in Analysis tab.
Notice that a successful results file attachment is indicated in the message field on the bottom left, as shown
in Figure 6-33.

Figure 6-33 Completion of attaching XDB file for post processing

Figure 6-34 Selecting Results tab for plotting mode shapes

5. Next, click on the Results tab and then click the Fringe/Deformation tool in the Quick Plot set (see
Figure 6-34).
6. In the Select Result Cases field, click on Mode 1 as shown in Figure 6-35.
7. In the Select Fringe Result field, click on Eigenvectors, Translational. In the Select Deformation
Result field, click on Eigenvectors, Translational. Click Apply. This generates the first mode shape
as shown in Figure 6-36.

Main Index
 110 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-35 Plotting first mode shape

Main Index
 Chapter 6: Rotordynamics with MSC Patran 111
Example Model

Figure 6-36 First mode shape for hollow rotor

8. This Image can be saved using following steps: Click on File and then Images. This open the Output
Tab as shown in Figure 6-37.
9. In the Output tab, select “Current Viewport” as Source and pick any desired output format from the
Image Format menu.

Figure 6-37 Steps for saving modeshape

10. Animate the mode by clicking the Animation Options button as shown in Figure 6-38.

Main Index
 112 MSC Nastran Rotordynamics User’s Guide
Example Model

11. In the Number of Frames field, specify 32 as shown and click Apply.
12. Click the Select Results button and then check Animate and click Apply. Click the No Graphics
Refresh button to stop the animation.

Figure 6-38 Steps involved in generating animation for modeshape

Repeat this procedure to investigate first-order forward and backward whirling bending modes 1
through 4, axial modes 5 and 6, second-order FW and BW bending modes 7 through 10, and
potentially further.
13. Patran animations can be saved for future usage in presentations by checking the MPEG option in
the animation box as shown in Figure 6-38.
Note that the main purpose of this exercise and Patran use in general is the configuration of analysis input
decks (i.e. preprocessing) and the visualization of results (i.e. postprocessing). Since the analysis preference
was chosen to be MSC Nastran in this exercise, an MSC Nastran Bulk Data File or *.bdf input deck was
written.
It is highly recommended that users investigate the input deck created by Patran and understand different
cards defined in the model. Details of each card are available in MSC Nastran’s Quick Reference Guide. A
snapshot Snapshot of the MSC Nastran input file is shown in Figure 6-39.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 113
Example Model

Figure 6-39 Input deck created by MSC Patran

Patran is extremely useful when working in the overall geometric context and orientation of large
subassemblies that contain rotors such as engines or entire systems such as airframes. At the same time, we
would like to emphasize MSC Nastran’s modularity that allows the user to go seamlessly between component,
subassembly, and system contexts.

Main Index
 114 MSC Nastran Rotordynamics User’s Guide
Example Model

Step 11: Generating Campbell Diagrams

Currently, MSC Patran does not include features to initiate generation of Campbell diagram in MSC
Nastran. This has to be done manually by modifying the input deck as shown in Figure 6-40. The
modifications required include:
a. Specifying two.output4 files which includes information about rotor speed and corresponding
eigenfrequencies.
b. CAMPBELL option in case control
c. CAMPBLL and DDVAL entries in bulk data section

Figure 6-40 Modification in input deck for generation of Campbell Diagram

When this modified input file is run in MSC Nastran, two additional files, namely campbell_freq.op4 and
campbell_rpm.op4 are generated. These files are required as input for generating a Campbell diagram in
Patran.
Here are the steps involved in generating a Campbell diagram.
1. Click on Utilities under the main menu Figure 6-41(a). And then click on Results and Campbell
Diagram. This will open a tab as shown in Figure 6-42(b).
2. In the Campbell Diagram Creation tab, click on Select Rotating speed file. In the pop-up window,
select the campbell_rpm.op4 file which was created during the Nastran run. Click OK to close the
window. (see Figure 6-43)
3. Similarly, click on Select Frequency File, and select the campbell_freq.op4 file.
4. Next, enter [0 1] in the Rotation Speed Multiplier to generate the reference line where rotor speed is
equal to the eigenfrequency. And finally, click on Apply to generate the Campbell diagram. The
Campbell Diagram generated for this particular problem is shown in Figure 6-44.

Main Index
 Chapter 6: Rotordynamics with MSC Patran 115
Example Model

Figure 6-41 (a) Generating Campbell Diagram in Patran

Main Index
 116 MSC Nastran Rotordynamics User’s Guide
Example Model

Figure 6-42 (b) Generating Campbell Diagram in Patran

Main Index
 Chapter 6: Rotordynamics with MSC Patran 117
Example Model

Figure 6-43 Attaching rotor speed and eigenfrequency files

Figure 6-44 Campbell Diagram

Main Index
 118 MSC Nastran Rotordynamics User’s Guide
Example Model

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 6-1
Input File Name Description
rug_6_1 Rotor model developed using Patran.
rug_6_2 Rotor model modified to generate campbell diagram.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors

Complex Eigenvalue
7 Analysis for Rotors


Complex Eigenvalue Analysis for Rotors 120
 Example 130

Summary for Equation of Motion 148

Main Index
 120 MSC Nastran Rotordynamics User’s Guide
Complex Eigenvalue Analysis for Rotors

Complex Eigenvalue Analysis for Rotors

Complex eigenvalue analysis can be used to determine whirl frequencies and critical speeds, and their
associated modeshapes. Complex eigenvalue analysis is available as a direct method (SOL 107), in which the
equations are of the same size as the number of physical variables. Complex eigenvalue analysis is also
available as a modal method (SOL 110), in which undamped modes are first computed and then used to
transform the matrices from physical to modal variables. Please refer to the Complex Eigenvalue Analysis (Ch. 14)
in the for more details on complex eigenvalue analysis.
A characteristic of the motion of rotors is that the rotor appears to "whirl" in a circular or elliptical motion
relative to the support structure. The whirl motion may be either in the direction of rotor rotation (forward
whirl) or against it (backward whirl), as described in Chapter 1. Rotor whirl can be at a frequency that is
different from the rotor spin rate.
Whirl frequencies are important in determining rotor stability. Although the whirling motion may not be at
the rotational speed, whirl modes may be unstable due to internal rotor damping or external forces due to
bearing and seals. Estimating the stability of the whirl modes is important for ensuring a system operates
smoothly. Critical speeds can be viewed as a special case of whirl frequencies: the rotor whirl is at the same
frequency as the rotor spin rate. Critical speeds are important for ensuring safe vibration levels.

Equations of motion
For complex eigenvalue analysis, final equation of motion in a fixed coordinate system, as derived in Chapter
2, is given by:

[ M ]{ g} + ([ BS ] + [ BR ] + Ω[G ]){ g } + ([ K ] + Ω[ K C ]R + Ω 2 [ K G ]) {g} = {FS }

The centrifugal stiffening matrix [KG] is important for 3D rotors where the model includes rotating
components away from the axis of rotation. For simplicity, this term is removed from the equations presented
in rest of the chapter.

Let the DOF vector be given by: { g } = { g a ( p )} pt where, {g a (p)} is the amplitude of motion (independent
of time) and p is complex eigenvalue for the model.
In order to determine whirl frequencies, the above equation is solved for p (complex eigenvalue) and {ga (p)}
(mode shape) with the rotor spin rate Ω specified by the user (asynchronous option). Here,
p = α + iω
where, α is the real component of the eigenvalue while ω is the imaginary component. Thus, ω represents the
oscillation frequency for the model and Re{ga (p)} represents the corresponding modeshape. For this case, the
damping coefficient produced as output in the complex eigenvalue analysis is given by:
Damping Coefficient (g)= -2α/ω
Thus, the system is considered unstable when (-2α/ω) < 0.
Critical speeds can be obtained by using the synchronous option is RGYRO as described in the previous
chapter.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 121
Complex Eigenvalue Analysis for Rotors

For Analysis in Fixed Reference Frame

1. Asynchronous mode (ASYNC option in RGYRO)
Equation of motion for asynchronous analysis to determine whirl frequencies of the system is obtained by
substituting {g} = {g a ( p)}e pt in the above equation:

(p 2
[ M ] + p ([ B S ] + [ B R ] + Ω [ G ] ) + ([ K ] + Ω [ K C ] R ) ) { g a ( p )} = 0

Above equation can be further expanded to include different options available for defining damping for rotor
and stator portion in the model. The final expression as included in MSC Nastran is given by:

 p 2[M ] + 
 
  ([ BS ] + α 1[ M S ] + α 2 [ K S ] + [ BH S ]) +  
   
 p  n  Ω j (Ω ref ) [G ] j + [ BR ] j + [ BH R ] j + α 1Rj [ M R ] + α 2 Rj [ K R ] j +   + 
    
   GR j 1 1  
   WR3 [ K R ] j + WR 4 [ K 4 R ] j + WRH [ KH R ] j    {g ( p )} = 0
j =1
 a
 
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] + 
 
 n    BRC  +  BH RC  + α 1Rj  M RC  + α 2 Rj  K RC  +   
       
   [ K ] + Ω (Ω )  
j j j

 j =1  R j j ref  GR 1 1 
    K C
 +  K 4 C
 +  KH C

R  j 
 WR3   j WR 4 
Rj
WRH 
R
  

where
[M] Total mass matrix for the model
i –1

p Complex eigenvalue of the model

[BS] Viscous damping from dampers in the stator portion
[MS] Mass matrix for the stator portion
[KS] Stiffness matrix for the stator portion
[BHS] Hybrid damping for the stator portion
α1, α2 Rayleigh damping parameters for stator portion
Ωref Rotor speed of the reference rotor (specified by the user)
Ωj Rotor speed for the jth rotor
[G] Gyroscopic matrix for the rotor
[BR] Viscous damping from dampers for the rotor
[MR] Mass matrix for the rotor

Main Index
 122 MSC Nastran Rotordynamics User’s Guide
Complex Eigenvalue Analysis for Rotors

[KR] Stiffness matrix for rotor

[BHR] Hybrid damping for the rotor
α1R, α2R Rayleigh damping parameters for the rotor
GR Structural damping for the rotor defined in RSPINR
[K4R] Material damping for the rotor part
[KHR] Hybrid damping for the rotor part
WR3, WR4, Parameters to convert structural damping to equivalent viscous damping.
WRH
G Structural damping for the stator portion
[K4S] Material damping for the stator portion
[KHS] Hybrid structural damping for stator
Circulation matrix due to [BR]
 BRC 

BHRC  Circulation matrix due to [BHR]

α1[ M RC ] + α 2 [ K RC ] Circulation matrix due to Rayleigh damping for rotors

GR Circulation matrix due to structural damping for rotors

 K RC 
WR3  

1 Circulation matrix due to structural damping for rotors

 K 4 CR 
WR 4 

1 Circulation matrix due to hybrid damping for rotor

 K H RC 
W RH 

{ga(p)} Modeshape corresponding to complex eigenvalue p

In the above equation, it is assumed that the parameters WRH, WR3, and WR4 are defined in the analysis
(either through PARAM or through RSPINR for each rotor) and not equal to zero. In case any of these
parameters are not defined, the damping defined using GR, GE and KH and corresponding circulation terms
are ignored in the analysis.

Synchronous Analysis
For the complex modes analysis with synchronous excitation, the excitation frequency is equal to the spin rate
of the reference rotor; that is, ω = Ω ref = Ω . In case multiple rotors are present in the analysis, the rotor
speed for each rotor is written as a linear function dependent on the reference rotor spin

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 123
Complex Eigenvalue Analysis for Rotors

rate: Ω ( Ω ref ) = α j + β j Ω ref . The scaling factors, αj and βj are determined from the relative spin rates
specified by the user. For this case, the equation of motion to be solved is obtained by substituting p = iΩ in
the previous equation:

(− Ω 2
[ M ] + i Ω ([ B S ] + [ B R ] + (α + β Ω )[ G ] ) + ([ K ] + (α + β Ω )[ K C ] R ) ) { g a ( Ω )} = 0

This simplifies to:

( − Ω ([ M ] − i β [ G ] ) + i Ω ([ B
2
S ] + [ B R ] + α [ G ] − i β [ K C ] R ) + ([ K ] + α [ K C ] R ) ){ g a ( Ω )} = 0

The above equation can be expanded based on different options available for defining damping for the stator
and for rotors. If parameters WR3, WR4, and WRH are defined in the analysis and not equal to zero, the
expanded version of the equation is given by:

 2
n  
 −Ω  [ M ] − i  β j [G ] j  + 
  j =1  
 
  ([ BS ] + α 1[ M S ] + α 2 [ K S ] + [ BH S ]) +  
   
    [ BR ] j + [ BH R ] j + α1Rj [ M R ] j + α 2 Rj [ K R ] j +    
       
α [G ] j +  GR  + 
   j 
j
[ K ] +
1
[ K 4 ] +
1
[ KH ] 
 iΩ n    
  +
R R R
 WR3 j
WR 4 j
WRH j

  
  j =1    BRC  +  BH RC  + α 1Rj  M RC  + α 2 Rj  K RC  +     g a (Ω) = 0
       
   −i β    
j j j j

   j  GR 1 1   
    
j
 K C
 +  K 4 C
 +  KH C
 j    

 WR3   j WR 4   j WRH 
R R R
    
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] + 
 
 n    BR  +  BH R  + α1Rj  M R  + α 2 Rj  K R  +  

C
 
C
 
C
 
C
 
 [K ] + α  j j j j  
  R j j  GR 1 1  
 j =1 
  K R  +
C
 K 4 R  +
C
 KH R   
C

   WR3 j WR 4 j WRH j  

Note that for the analysis in fixed reference frame, the synchronous complex eigenvalue analysis produces
both forward and backward whirl modes. Identification of these modes can be done using Campbell diagram
or by post processing the eigenvectors produced as part of output.

For analysis in rotating reference frame

Equation of motion for complex eigenvalue analysis when the problem is analyzed in the rotating reference
frame is given by (as derived in Chapter 2):

[ M ] { g} + ([ B S ] + [ B R ] + 2 Ω [ C ] ) { g } + ([ K ] − Ω 2 [ Z ] + Ω 2 [ K G ] + Ω [ K C ] S ) { g } = { FR }

Main Index
 124 MSC Nastran Rotordynamics User’s Guide
Complex Eigenvalue Analysis for Rotors

Substituting {g } = {g a ( p )}e pt and {FR} = 0 in the above equation

( p [ M ] + p ([ B ] + [ B
2
S R )
] + 2 Ω[C ] ) + ([ K ] + Ω[ K C ]S − Ω 2 [ Z ] + Ω 2 [ K G ] ) { g a ( p )} = 0

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 125
Complex Eigenvalue Analysis for Rotors

Asynchronous Analysis
Expanded version of the above equation for asynchronous analysis is given by:

 p 2[M ] + 
 
  ([ BS ] + α 1[ M S ] + α 2 [ K S ] + [ BH S ])  
  n  
 p+ +
   (
j =1
2Ω [C ] j + [ BR ] j + [ BH R ] j + α 1Rj [ M R ] j + α 2 Rj [ K R ] j 
 
)

 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] +  { g a ( p )} = 0
 
(
 Ω  BS  +  BH S  + α 1  M S  + α 2  K S  +
C C C C
) 
 n 

 ([ K R ] j (1 + iGR j ) + i[ K 4 R ] j + i[ KH R ] j − Ω [ Z ] j + Ω [ K G ] j ) 
2 2

j =1
 
 

Note that in case multiple rotors are present in the analysis, they are constrained to have the same rotational
speed and axis of rotation for the analysis in rotating reference frame.

Synchronous Analysis
In rotating reference frame, different sets of equations exist to determine forward whirl and backward whirl
critical frequencies. And thus, separate analysis need to be performed to obtain the backward whirl critical
frequencies with the parameter WHIRLOPT turned on.
In order to obtain the forward whirling critical speeds, the criteria used is: ω = 0 or (p = 0).
Substituting this expression in the simplified equation, the following equation is obtained:

([ K ] + Ω [ K C ] S − Ω 2 [ Z ] + Ω 2 [ K G ] ) { g a ( Ω )} = 0

This can be written in standard quadratic form as:

( − Ω ([ Z ] − [ K ] ) + i Ω ( − i [ K
2
G C ] S ) + [ K ] ) { g a ( Ω )} = 0

Expanding the above equation to include damping and circulation effects:

 2 n 
 −Ω  ([Z ] j − [ KG ] j ) 
 j =1

 (
 +iΩ(−i)  BC  +  BH C  + α1  M C  + α 2  K C  +
 S  S  S  S )  { g (Ω)} = 0
 a
 n 
 (1 + iG)[ KS ] + i[ K 4S ] + i[ KH S ] +  ([ K R ] j (1 + iGR j ) + i[ K 4R ] j + i[ KH R ] j ) 
 j =1 

Main Index
 126 MSC Nastran Rotordynamics User’s Guide
Complex Eigenvalue Analysis for Rotors

Note that this equation does not include the effect of viscous damping defined in the fixed or rotating
reference frame. Also, although the above equation does not include mass matrix explicitly, the centripetal
matrix [Z] and differential stiffness matrix [KG] are dependent on rotor mass distribution.

Backward whirl critical frequencies can be obtained by using the following substitution:
ω = 2Ω, that is, p = i(2Ω)
Substituting this expression in the above equation, the equation to determine backward whirl critical
frequencies is given by:

 2   
( )
n
 −Ω  4[ M ] +  − 4 i [C ] j + [ Z ] j − [ K G ] j  + 
  j =1  
 
  2 ([ B S ] + α 1 [ M S ] + α 2 [ K S ] + [ BH S ]) +  
  n  
 iΩ   
  (
2 [ B R ] j + [ BH R ] j + α 1Rj [ M R ] j + α 2 Rj [ K R ] j ) +
  { g a ( Ω )} = 0
  j =1
  ( − i )  B SC  +  BH SC  + α 1  M SC  + α 2  K SC 
( )  

           
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] + 
 n 
 
  ([ K R ] j (1 + iGR j ) + i[ K 4 R ] j + i[ KH R ] j ) 
 j =1 

Solvers available for complex eigenvalue analysis

MSC Nastran provides different options for solver to be used in complex eigenvalue analysis. In SOL 107
analysis, CMETHOD in case control selects parameters for complex eigenvalue extraction. Actual parameters
to be used in the analysis are specified using the EIGC card in bulk data. For example:
In case Control:
CMETHOD = 100
In Bulk Data:
EIGC 100 CLAN 20
For rotordynamic analysis, where the matrices can be complex and skew symmetric, use of HESS (Upper
Hessenberg) and CLAN (Complex Lanczos) method is recommended. More details about these solvers is
given in MSC Nastran Quick Reference Guide and MSC Nastran’s Numerical Methods User’s Guide.
The HESS method is generally more reliable and appropriate for small size problems only. It can take
significant computational time for moderate or large size problems. For large size problem, use of appropriate
reduction methods is encouraged to reduce the size of the problem in case the user needs to use HESS option.
The CLAN method is better suited for moderate or large problem due to its computational efficiency.

Pardiso Solver
In MSC Nastran, Intel MKL Pardiso direct solver is introduced that can help in achieving upto 80%
reduction in total elapsed time versus the default MSC solver. Further details about the Pardiso solver is

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 127
Complex Eigenvalue Analysis for Rotors

provided in the MSC Nastran Release Guide. Note that Intel MKL Pardiso is only available with the Lanczos
(CLAN) method for eigenfrequency analyses.
You are required to use the SPARSESOLVER command in the executive section to indicate that the new
Pardiso solver should be used in the simulation. The following is example of choosing the new sparse direct
solver using the keywords PRDLU (for unsymmetric matrices),
SOL 107
SPARSESOLVER CEAD (FACTMETH=PRDLU,ORDMETH=PRDSMPMS)
CEND

Mode Tracking
In V2018, two different approaches were implemented for mode tracking while generating Campbell
diagram. In both the methods, only the complex frequencies with a positive imaginary part are considered
for mode tracking.

Numerical Approach
This method is used for mode tracking once the eigenvalues have been extracted for all the rotor speeds
defined for analysis using CAMPBLL and DDVAL bulk data entries. For each rotor speed, the second
derivative is computed using the central difference approach.
For any rotor speed (Ωi), if the kth mode is tracked for the previous two rotor speeds (Ωi-1 and Ωi-2), then
the second derivative is computed for all the N modes using:

f j (Ωi ) + f k (Ωi − 2 ) − 2 f k (Ωi −1 )

f j '' =
(Ωi − Ωi −1 )(Ωi − 2 − Ωi −1 )
where, 1<j<N, N being the number of eigenvalues with positive imaginary part. Of all the frequencies at rotor
speed Ωi, the one which results in least absolute value for second derivative ( f '' ) is matched to the kth mode.
For the second rotor speed in the analysis, where the results for Ωi-2 are not available, the first derivative is
used to check the continuity. This method is the default option for mode tracking due to its robustness.

Eigenvector-Based Approach
Unlike the numerical approach, here both left and right eigenvectors associated with complex eigenvalues are
used for mode tracking. The method implemented is described in detail in Reference 2. As described in the
paper, Normalized Cross Complex Orthogonality (NC2O) parameter is computed for each pair frequencies
for two rotor speeds. In V2018, the eigenvector based mode tracking approach is available for analysis in
FIXED reference frame only when HESS option is used in EIGC for eigenvalue extraction. Due to
limitations of the HESS option when working with large size models without a reduction, the use of SOL
110 or a superelement based reduction approach is recommended for analyzing large rotor models.
Additional features added to make the eigenvector-based mode tracking more robust are described below:
Parameter CORU: Ideally, for tracked pair of modes, the value of parameter NC2O is very close to 1.
However, for complex models involving large stator component and dynamic reduction, this value can be

Main Index
 128 MSC Nastran Rotordynamics User’s Guide
Complex Eigenvalue Analysis for Rotors

significantly lower. Thus, the default value is set to 0.7. Based on the model complexity and NC2O matrix
obtained for two rotor speeds, the user may wish to change this value.
Parameter SWITR: In case of mode tracking failure at a particular rotor speed, user has the option to
continue mode tracking for rest of the rotors speed by:
SWITR = 1: Updating modes corresponding to rotor speed with failed mode tracking, or
SWITR = 0: Using the modes from last rotor speed with successful mode tracking. (Default)
In cases where the Campbell diagram includes multiple mode switching and/or there are new modes entering
the analysis, use of SWITR=1 is recommended.
Parameter NUMMODE: As mentioned earlier, only the complex eigenvalues with positive imaginary part
are considered for mode tracking analysis. As a result, the number of eigenvalues available for mode tracking
may not be same at all the rotor speeds considered in the analysis. This may lead to mode tracking failure.
One way to avoid this issue is to pick NUMMODE to be less than half of the number of complex eigenvalues
extracted using EIGC bulk data entry. Thus, the analysis uses only the first NUMMODE frequencies with
positive imaginary part for mode tracking at each rotor speed.
Parameter PRTCOR: When this parameter is turned to 1, the code prints NC2O matrix corresponding to
each rotor speed in the F06 file. This information can be very useful in identifying reasons for mode tracking
failure.
Additional guidelines and limitations for mode tracking are described below:
 CAMPBLL option is not supported in SOL 200 and SOL 400 for ANALYSIS = DCEIG/MCEIG.
 Only the modes with positive imaginary part are used for mode tracking (to avoid repeated
eigenvalues.)
 Eigenvector based mode tracking is not supported for analysis in rotating reference frame.
 For both approaches, damping ratio is also tracked while tracking eigenfrequencies. The
results produced from mode tracking can be extracted in OP4 using following assign
statements:
$ Extract results for numerical mode tracking
ASSIGN OUTPUT4='freq_nmt.op4',UNIT=71,FORM=FORMATTED,DELETE $ MODE FREQS
ASSIGN OUTPUT4='rot_nmt.op4',UNIT=72,FORM=FORMATTED,DELETE $ ROTOR SPEED
ASSIGN OUTPUT4='dmp_nmt.op4',UNIT=73,FORM=FORMATTED,DELETE $ DAMPING RATIO
$ Extract results for eigenvector based mode tracking
ASSIGN OUTPUT4='freq_evmt.op4',UNIT=81,FORM=FORMATTED,DELETE $ MODE FREQS
ASSIGN OUTPUT4='rot_evmt.op4',UNIT=82,FORM=FORMATTED,DELETE $ ROTOR SPEED
ASSIGN OUTPUT4='dmp_evmt.op4',UNIT=83,FORM=FORMATTED,DELETE $ DAMPING RATIO

These OP4 files can be imported in MSC PATRAN to produce Campbell Diagram.

Mode Identification
Complex eigenvectors for rotors demonstrate whirl motion. It can be forward whirl or backward whirl in
nature. Using complex eigenvectors, it is possible to determine the nature of the whirl motion (Reference 2).

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 129
Complex Eigenvalue Analysis for Rotors

This approach makes use of complex displacement in the two directions orthogonal to rotor spin axis. As a
result, this method may not work well when rotors are reduced using normal modes.
For a rotor spinning about the Z axis with the X and Y axis being orthogonal to the spin axis, let the complex
displacement be given by:

u x = u x , R + iu x , I
u y = u y , R + iu y , I

Whirl motion at each rotor grid can be expressed as a combination of forward whirl and backward
whirl motion. Using the approach described in Ref 2, it is possible to determine the magnitude of
forward whirl and backward whirl radii. Final expression for these radii is given by:

1
Forward whirl radius: | w f |= (u x , R + u y , I ) 2 + (−u x , I + u y , R )2
2
1
Backward whirl radius: | wb |= (u x , R − u y , I ) 2 + (u x , I + u y , R ) 2
2
Whirl direction is determined by looking at the difference between the two radii:

| wb | − | w f |
w=
| wb | + | w f |
If (w<0), then it represents forward whirl motion, while if (w>0) then it represents backward whirl motion.
In case w=0, it indicates pure bending motion. This analysis can be done for each rotor grid to determine the
whirl direction.
In the current implementation, it is possible to extract values of w, wf, and wb for each rotor grid, if desired.
In V2018, the mode identification approach is available for 1D rotors defined using ROTORG entry.

Mode Identification
In order to identify a mode, the following values are determined using all the rotor grids:
max(w),min(w)
Based on the value of these parameters, the whirl mode is identified using following rules:
 |wmax| > 0.8 and |wmin| < 0.005 : Backward Whirl Mode (BW)
 |wmax| > 0.1 and |wmin| < 0.005 : Partial Backward Whirl Mode (BP)
 |wmax| < 0.005 and |wmin| > 0.8 : Forward Whirl Mode (FW)
 |wmax| < 0.005 and |wmin| > 0.1 : Partial Forward Whirl Mode (FP)
 |wmax| > 0.005 and |wmin| > 0.1 : Mixed Whirl Mode (MX)
 |wmax| > 0.1 and |wmin| > 0.005 : Mixed Whirl Mode (MX)

Main Index
 130 MSC Nastran Rotordynamics User’s Guide
Example

If any of the above criterion are not satisfied, code looks for the presence of axial (AX) and torsion
(TR) modes.
Consider following example where the value of parameter “w” is obtained for four different modes
for all the rotor grids.

W Mode A Mode B Mode C Mode D

Grid 1 0.0008 -0.0027 -0.0039 0.0062
Grid 2 0.3812 -0.4381 -0.0806 0.1316
Grid 3 0.5265 -0.5942 -0.0874 0.1313
Grid 4 0.6639 -0.7351 -0.0788 0.0733
Grid 5 0.7909 -0.8579 -0.0360 -0.0054
Grid 6 0.9998 -0.9987 -0.0535 0.1426
Grid 7 0.8327 -0.8719 -0.2335 -0.0047
Grid 8 0.6530 -0.7217 -0.4142 0.1001
Grid 9 0.4678 -0.5433 -0.3714 0.2724
Grid 10 0.0005 -0.0097 -0.0323 0.0702
max(w) 0.9998 -0.0027 -0.0039 0.2724
min (w) 0.0005 -0.9987 -0.4142 -0.0054
Mode Type BW FW FP MX

For Mode A, |wmax|=0.9998 and |wmin|=0.0005. Thus, according to the first criteria, this mode
is identified as a Backward Whirl (BW) mode.
For Mode B, |wmax|=0.0027 and |wmin|=0.9987. Thus, according to the third criteria, this mode
is identified as a Forward Whirl (FW) mode.
For Mode C, |wmax|=0.0039 and |wmin|=0.4142. Thus, according to the fourth criteria, this mode
is identified as a partial Forward Whirl (FP) mode.
For Mode D, |wmax|=0.2724 and |wmin|=0.0054. Thus, according to the sixth criteria, this mode
is identified as a Mixed Whirl (MX) mode.

Example
In this section, above equations are used to determine whirl and critical frequencies for a representative rotor
model. The analysis is performed in both fixed and rotating coordinate systems. For the purpose of
comparison, same rotor is modeled using 1-D elements (ROTORG), axisymmetric harmonic elements
(ROTORAX), and solid elements (ROTOR). For each case, first the Campbell diagram is obtained using
asynchronous analysis, and then in the next step, critical frequencies are obtained using synchronous analysis.
The finite element models used in the analysis are shown in Figure 7-1. Since the rotor model is symmetric,

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 131
Example

the problem can be analyzed in both fixed and rotating reference frame. When analyzed in rotating reference
frame, the user needs to ensure that the stator portion of the model is symmetric.

Figure 7-1 NM Rotor modeled using three different elements

Analysis in Fixed Reference Frame

First, the analysis is performed in fixed reference frame, and Campbell diagram and critical frequencies are
obtained for the above rotor models. For the first case, the rotor is supported with symmetric springs at two
of the nodes.
Case 1: Spring stiffness used: KXX = KYY = 4.38x107 N/m2
The Campbell diagram for this case is shown in Figure 7-2. In case of 1-D modeling, two different modeling
techniques are applied using CBAR and CBEAM elements. It can be seen here that at 0 RPM, natural
frequencies predicted by different models are very close to each other. Results for variation of whirl frequency
with rotor speed show that all the models considered show good match for lower frequencies; however, results
start to separate for higher frequencies.

Main Index
 132 MSC Nastran Rotordynamics User’s Guide
Example

Figure 7-2 Campbell Diagram in the fixed system

Critical frequencies can be obtained using the Campbell diagram by determining the point of intersection of
whirl frequency line with 45 degree line (dotted black line) as shown in Figure 7-2. Critical frequencies can
also be obtained by performing synchronous complex eigenvalue analysis by selecting SYNC option in
RGYRO. Results obtained for different models are listed in . Mode shapes for the first three forward whirl
bending critical frequencies is shown in Figure 7-3. The results show that the mode shapes captured by beam
model and solid element model for this simple rotor model look similar.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 133
Example

Table 7-1 Critical frequencies in fixed reference frame

Mode type Frequencies (Hz)
Bar Model Beam Model Axi Model Solid Rotor
1st Bending Backward Whirl 259.1 257.3 257.1 257.63
1st Bending Forward Whirl 282.5 284.7 279.9 280.42
2nd Bending Backward Whirl 794.4 767.6 744.6 746.93
2nd Bending Forward Whirl 795.7 820.1 800.1 804.19
Torsion Mode 1097.9 1025.4 1005.79
3rd Bending Backward Whirl 1122.4 1088.7 1117.5 1130.72
3rd Bending Forward Whirl 1529.2 1628.6 1528.5 1457.38

Beam Model Solid Element Model

Bending
Mode 1

Bending
Mode 2

Bending
Mode 3

Figure 7-3 Bending mode shapes

In the first test case, bearings stiffness used at the ends did not include cross-stiffness. It is possible to include
the effects of cross-stiffness in springs indirectly using CBUSH or DMIG entry. In case KXX = KYY and KXY=
KYX, it is possible to define cross stiffness using CBUSH. However, for more general cases, use of DMIG
entry is recommended. In this modified case, following values are used for spring stiffnesses in the analysis.
Case 2: KXX = KYY = 3.5025 x 107 N/m2

Main Index
 134 MSC Nastran Rotordynamics User’s Guide
Example

KXY = KYX = 0.8755 x 107 N/m2

Both the available approaches are shown in Figure 7-4. The results obtained here compare well with those
obtained by Geradin et al. (Geradin and Kill, 1984), see Table 7-2.

Figure 7-4 Approach to define cross stiffness for bearings

Table 7-2 Critical Frequencies with modified support

Table 7-3
Mode type Frequencies (Hz)
Bar Model Beam Model Axi Model Solid Rotor
1st Bending Backward Whirl 232.4 231.6 231.1 232.25
1st Bending Forward Whirl 273.1 274.1 270.8 271.34
2nd Bending Backward Whirl 657.6 655.4 643.1 647.52
2nd Bending Forward Whirl 794.8 793.6 772.7 775.96
Torsion Mode 1036.8 1006.1 1004.34
3rd Bending Backward Whirl 1097.9 1025.4 1026.6 1037.20
3rd Bending Forward Whirl 1472.1 1580.2 1481.9 1456.01

In the next case, the damping is introduced in the fixed system through the spring supports at the end (also
referred as external damping).
Case 3: KXX = KYY = 4.38x107 N/m2

BXX = BYY = 2.627x103 N/m

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 135
Example

Same damping can also be introduced in the rotating system by defining spring supports on grid points which
are part of the rotor, as shown in Figure 7-5 (also referred as internal damping). When internal damping is
present in the system, circulation terms are generated and included in equation of motion.

Figure 7-5 CBUSH definition for non-rotating and rotating portion

(solid line: external damping, dotted line: internal damping)

Figure 7-6 Campbell diagram with damping in stator

Main Index
 136 MSC Nastran Rotordynamics User’s Guide
Example

The effect of internal and external damping on complex eigenfrequencies and damping ratios is shown in
Figure 7-6 and Figure 7-7, respectively. The Campbell diagram presented in Figure 7-6 shows small effect of
damping on eigenfrequencies. However, the damping ratios shown in Figure 7-7 show very different results for
the internal and external damping cases. In case of external damping (defined in nonrotating portion),
damping results in positive damping ratio indicating increase in system stability. The results also show
increase/decrease in damping ratio with rotor speed, depending on the whirl mode. However, for the case
with internal damping, forward whirl modes demonstrates significant decrease in damping ratio with increase
in rotor speed and highlight that these modes may become unstable at certain rotor speed.

Figure 7-7 Effect of internal and external damping

Mode Tracking Example

Mode tracking feature implemented in V2018 is demonstrated using the Nelson-McVaugh rotor model
shown earlier in Figure 7-1.
The range of rotor speed for analysis is specified using the DDVAL entry, as described below.
DDVAL 22 8000.0 THRU 10. BY -400.
CAMPBLL 15 SPEED 22 FREQ
1 0.85 0 1

Numerical Mode Tracking

Result obtained using numerical mode tracking approach is shown in Figure 7-8

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 137
Example

Figure 7-8 Numerical Mode tracking results for 3D rotor

Eigenvector based mode tracking

Figure 7-9 Eigenvector based Mode tracking results for 3D rotor

The Campbell diagram obtained using eigenvector based mode tracking is presented in Figure 7-9. It matches
the Campbell diagram shown in Figure 7-8. Here, it appears that mode tracking has failed for mode 5 and 6
at 144000 RPM (2400 Hz) speed. This can be observed in the output provided in F06 file for this particular
rotor speed.
The mode tracking information produced in F06 file when PRTCOR parameter is turned to 1 is described
below:

Main Index
 138 MSC Nastran Rotordynamics User’s Guide
Example

The modal correlation matrix is the value of NC2O parameter for each pair of mode at two rotor speeds
where frequencies are being tracked.
MODAL CORRELATION MATRIX (PRINTED ROW-WISE)
0.9982 0 0.0001044 0 0.0004881 0.0001539 0 0 0 0.0001434 0
0 0.9984 0 0 0 0 0 0.000197 0 0 0
0 0 0.9972 0 0.001529 0.0004959 0 0 0.0001572 0.0001493 0
0 0 0 0.9994 0 0 0 0.0001159 0 0 0.0001963
0 0 0.0002733 0 0.6113 0.3881 0 0 0 0 0
0.0005172 0 0.001699 0 0.3854 0.6108 0 0 0.0002347 0.0005181 0
0 0 0 0 0 0 0.9999 0 0 0 0
0 0.0002442 0 0.0001368 0 0 0 0.9942 0 0 0.004252
0 0 0.0001635 0 0.0001765 0 0 0 0.9992 0.0002286 0
0 0 0 0.0001813 0 0 0 0.004441 0 0 0.9939
0.0001403 0 0.0001281 0 0.0004396 0.0001396 0 0 0.0002357 0.9981 0

Each term of the NC2O matrix is compared to CORU parameter, and it is turned to 1 if its greater than
CORU and turned to 0 if it is less than CORU.
MODAL CORRELATION MATRIX AFTER FILTER
1 0 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 1 0

MODE TRACKING FAILED FOR MODE 5

MODE TRACKING FAILED FOR MODE 6

Based on the result obtained above, it is determined if the modes have switched or if modes have failed to
track.
MAPPING WITH RESPECT TO PREVIOUS ROTOR SPEED
1 1
2 2
3 3
4 4
5 5 MODE FAILED
6 6 MODE FAILED
7 7
8 8
9 9
10 11 MODE SWITCHED
11 10 MODE SWITCHED

For this particular case, mode tracking failure is due to two modes being very close to each other at a particular
rotor speed. It is possible to avoid this mode tracking failure by excluding this rotor speed from the analysis.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 139
Example

DDVAL 22 8000. 7600. 7200. 6800. 6400. 6000. 5600. +

+ 5200. 4800. 4400. 4000. 3600. 3200. 2800. 2000.
+ 1600. 1200. 800. 400. 10.
Results obtained for this case are given in the following figure

Figure 7-10 Revised Mode Tracking Results

Here, the modal correlation matrix for rotor speed of 2800 Hz can be visualized as:

Figure 7-11 Modal Correlation

Main Index
 140 MSC Nastran Rotordynamics User’s Guide
Example

MAPPING WITH RESPECT TO PREVIOUS ROTOR SPEED

1 1
2 2
3 3
4 4
5 6 MODE SWITCHED
6 5 MODE SWITCHED
7 7
8 8
9 9
10 11 MODE SWITCHED
11 10 MODE SWITCHED

Mode Identification Example

The test case used to demonstrate mode identification feature is shown in Figure 7-11. It consists of 9 1D
beam elements defined using 10 grids. Its supported at the ends by CBUSH elements and has 5 additional
concentrated mass elements.

Figure 7-12 Test Case for Mode Identification

The Campbell diagram for this case and the mixed whirl modes identified are shown in Figure 7-12. In the
mixed whirl modes identified for this model, they appear at the intersection of forward and backward whirl
modes.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 141
Example

Figure 7-13 Mixed Whirl Modes identified from the analysis

According to the results presented in Campbell Diagram, one of the mixed whirl mode appears at 5000 RPM.
The value of parameter “w” for the mixed whirl mode at 5000 RPM is presented in Table 7-3. It can be seen
here that nodes 5 and 7 have forward whirl motion while the remaining nodes have backward whirl motion.
The corresponding mode shape is shown in Figure 7-13. The mixed whirl nature of the mode can be seen in
the animated mode shape.

Table 7-4 Value of whirl identification

parameter
Grid W
Node 1 0.0062
Node 2 0.1316
Node 3 0.1313
Node 4 0.0733
Node 5 -0.0054
Node 6 0.1426
Node 7 -0.0047
Node 8 0.1001
Node 9 0.2724
Node 10 0.0702

Main Index
 142 MSC Nastran Rotordynamics User’s Guide
Example

Figure 7-14 Mode shape for the whirl mode

Analysis in Rotating Reference Frame

The rotor model developed using solid elements can also be analyzed in the rotating coordinate system using
the ROTOR entry. The Campbell diagram obtained for this case is shown in Figure 7-15. The Campbell
diagram for analysis in rotating reference frame looks different from that obtained from analysis in the fixed
reference frame (see Figure 7-2). For the rotating reference frame Campbell diagram, the forward whirl critical
frequencies are identified by looking at the point of intersection of whirl frequency line with X axis (find Ω
such that ω = 0), while backward whirl frequencies are obtained by determining the point of intersection of
frequency lines with line corresponding to twice the rotor speed (ω = 2Ω shown in black, starting from
origin).
As in the fixed reference frame analysis, critical frequencies can also be obtained directly using the
synchronous analysis option in RGYRO for SOL 107. However, forward and backward critical frequencies
are obtained using two different analyses. Backward whirl critical frequency requires the WHIRLOPT
parameter to be listed as “BWD.” Results obtained using synchronous analysis a4rre shown in Table 7-5

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 143
Example

Figure 7-15 Campbell diagram in rotating reference frame

Table 7-5 Critical speeds in rotating reference frame

3D Rotor (Hz) 3D Rotor (RPM)
FWD Whirl BWD Whirl FWD Whirl BWD Whirl
1st Bending 274.29 252.59 16458 15155
2nd Bending 790.63 731.79 47438 43907
Torsion 1006.31 60379

Use of Whirl Parameter for Asynchronous Analysis

In complex eigenvalue analysis, it is sometimes desirable to specify whirl speed instead of WR3 parameter.
The whirl damping parameter (WR3WRL) is defined in the RGYRO entry. When this parameter is defined
in asynchronous complex eigenvalue analysis, the structural damping in rotors is included in the analysis as
viscous damping matrix using the following expression:

[ K R ]1 GR1 [ K R ]2 GR2
[ BR ] = +
| Ω 1 ( Ω ref ) − WR 3WRL | | Ω 2 ( Ω ref ) − WR 3WRL |

where, Ω1(Ω_ref ) and Ω2(Ω_ref ) are user-specified rotor speeds for Rotor 1 and Rotor 2 respectively.
For this rotor damping terms, corresponding circulation matrices are also generated and included in the
analysis.

Main Index
 144 MSC Nastran Rotordynamics User’s Guide
Example

Analysis in Fixed and Rotating Reference Frame

The main criteria used in determining whether to perform analysis in fixed or rotating reference frame is
dependent on the symmetry present in the rotor. In some cases, when both the rotor and the stator portion
of the model are symmetric, it is possible to analyze the same model in both fixed and rotating coordinate
reference frame. Based on the element implementation, 1-D elements like CBEAM and CBAR are better
suited for analysis in fixed system whereas higher order elements like shell and solid elements are better suited
for analysis in the rotating reference frame. For a simple case shown here, analysis is performed in both
rotating and fixed reference frame and results obtained for Campbell diagram and critical frequencies are
compared.

Relation between frequencies

Consider a simplified case where the analysis in fixed reference frame results in purely imaginary complex
conjugate pair of eigenvalues: p = ± iω . For this complex eigenvalue, corresponding response is of the form:

 
 u   u 0 cos ωt + φ u 
  =  
 v   v 0 cos ωt + φ v 
 

where u and v are lateral rotor displacement perpendicular to the axis of rotation and φu and φv are
corresponding phase differences. If Ω is the rotor speed, the corresponding mode shape in the rotating
reference is given by:

 
 u  cos ( Ωt ) sin ( Ωt )  u 0 cos ( ωt + φ u ) 
  =  
 v  – sin ( Ωt ) cos ( Ωt )  v 0 cos ( ωt + φ v ) 
 

Using the sum and difference formulae for sine and cosine, it can be shown that the frequencies ω ± Ω appear
in the solution. For example, the expression for u is given by:

This shows that the eigenfrequencies σ in the fixed reference frame are transformed to frequencies ω ± Ω in
the rotating reference frame. For the purpose of demonstration, consider a simple Jeffcott rotor model shown
in Figure 7-16. The rotor shaft is considered to be massless and rigid. The disk is attached at the center of the
shaft.

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 145
Example

Figure 7-16 Simplified rotor model

The Campbell diagram for this case is shown in Figure 7-17. Here, mode 1 and mode 2 correspond to the
translational motion whereas the mode 3 and mode 4 correspond to rotation of the shaft. The variation of
these modes with rotor speed is different in fixed and rotating coordinate reference frame. However, for this
simple case, it is possible to obtain the Campbell diagram in a rotating coordinate reference frame using the
Campbell diagram in a fixed reference frame by using the following expressions:

ω1 = ω1 + Ω : First backward whirl mode

ω2 = ω2 − Ω : First forward whirl mode

ω 3 = ω 3 + Ω : Second backward whirl mode

ω4 = ω4 − Ω : Second forward whirl mode

Results obtained using this transformation are compared with those obtained in the rotating reference frame
analysis in Figure 7-18. It can be seen here that both the set of results match very well.

Main Index
 146 MSC Nastran Rotordynamics User’s Guide
Example

Figure 7-17 Campbell Diagram in Fixed and Rotating System

Figure 7-18 Campbell Diagram in Rotating System

Additional Options for Damping

Additional options are introduced in MSC Nastran to control damping applied to rotor and stator portion
of the model. In order to implement these changes, an additional parameter called “RDBOTH” was

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 147
Example

introduced in the MDLPRM entry. The parameter is allowed to take values of 1, 2, 4, 8, or any combination
of it. Default value for this parameter is 0.

Effect of MDLPRM, RDBOTH, 1

1. Rayleigh damping specified in the model through “PARAM, ALPHA1” and “PARAM, ALPHA2” is
applied to the complete model (both stator and rotor).
2. Rayleigh damping specified through “ALPHAR1” and “ALPHAR2” in RSPINR/RSPINT is set to
0.0.
3. Rayleigh damping is also applied to “User supplied system” matrices that are supplied to MSC Nastran
via TF, K2PP, M2PP, and B2PP Nastran input

Modified equation of motion when “MDLPRM, RDBOTH, 1” is used in the analysis is given by:

 
 
 [ M ]{ g(t )} + 
  
( )
n
  ([ BS ] + α 1[ M ] + α 2 [ K ] + [ BH ]) +  Ω j ( Ω ref ) [G ] + [ BR ]  { g (t )} +  = {F (t )}
 j =1
j j
 
 
 (1 + iG )[ K ] + i[ K 4] + i[ KH ] +  [ K ] + Ω (Ω )   B C  + GR  K C    
n

 S   R j
j =1 
j ref   R  j
 WR 3 
R  j 



Note that here Rayleigh Damping parameters, α1 and α2, defined for the non-rotating component are
applied to the mass and stiffness matrix of the complete model (including rotors).

Effect of MDLPRM, RDBOTH, 2

When the parameter RDBOTH is assigned a value of “2”, the circulation terms are ignored in the analysis.
In this case, no distinction is made between damping applied to rotor and stator and both are included as
part of [B] matrix. Modified equation of motion when “MDLPRM, RDBOTH, 2” is used in the analysis is given
by:

 
 
 [ M ]{ g(t )} + 
  ([ B ] + α 1[ M ] + α 2 [ K ] + [ BH ]) +  
 S S S S
 
   Ω (Ω ) [G ] + [ B ] + [ BH ] + α 1 [ M ] + α 2 [ K ] +   
 n    g (t ) +  = {F (t )}
j ref j R j R j Rj R Rj R j

    GR j 1 1  
  j =1  [KR ]j + [ K 4R ] j + [ KH R ] j  
   WR 3 WR 4 WRH  
 n 
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] +  ([ K R ] j ) 
 j =1 

Main Index
 148 MSC Nastran Rotordynamics User’s Guide
Summary for Equation of Motion

Effect of differential stiffness

Figure 7-19 Hollow rotor model

Effects of stress stiffening are important for 3-D rotors, especially when the model includes large flexible
structure away from the axis of rotating, high rotational speeds, and/or large diameter blades. In order to
demonstrate its effect, a hollow shell rotor model is analyzed in rotating reference frame as shown in
Figure 7-19. Results shown in Figure 7-20 highlight the increase in stiffness with increase in rotor speed. The
flag for generating differential stiffness can be turned on by one of the following options:

MDLPRM RDBOTH 4
OR
MDLPRM RDBOTH 8

Figure 7-20 Effect of stress stiffness in rotating reference frame

Solid Line: without stress stiffening, Dotted Lines: with stress stiffening

Summary for Equation of Motion

Equations of motion for complex eigenvalue analysis are converted to the standard second order system prior
to solving them. Standard second order equation used to determine complex eigenvalues in MSC Nastran is
given by:

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 149
Summary for Equation of Motion

where [M]eq, [B]eq and [K]eq are equivalent mass, damping, and stiffness matrices, respectively, and these
matrices can be real or complex. In rotordynamic analysis, these matrices need not be symmetric.
For the asynchronous and synchronous analysis in fixed and rotating reference frame, equivalent matrices are
listed in Table 7-6.

Table 7-6 Coefficients for second order system of equations

Fixed Reference Frame

Table 7-7
[M]eq [B]eq [K]eq
ASYNC [M] [B]S + [B]R + Ω[G] [K]+Ω[KC]R
SYNC [M]-iβ[G] [B]S + [B]R + α[G] -iβ[KC]R [K]+ α[KC]R

Rotating Reference Frame

Table 7-8
[M]eq [B]eq [K]eq
ASYNC [M] [B]S +[B]R+2Ω[C] [K]+Ω[KC]S – Ω2[Z] + Ω2[KG]
SYNC (FWD) [Z] - [KG] -i[KC]S [K]
SYNC (BWD) 4[M] +4[C]+[Z]-[KG] 2[BS]+2[BR]-i[KC]S [K]

Main Index
 150 MSC Nastran Rotordynamics User’s Guide
Summary for Equation of Motion

Damping and circulation terms used for analysis in Fixed Reference Frame

Table 7-9
Damping Circulation
Solution: ASYNC option (PARAM WR3, WR4 and WRH defined)

   
 [ BR ] + α1R [ M R ] + α 2 R [ K R ] +   [ BRC ] + α 1R [ M RC ] + α 2 R [ K RC ] + 
   
iω 
GR
[KR ] +
1
[ K 4R ] +  Ω ref  GR [ K C ] + 1 [ K 4C ] + 
 WR3 WR 4   WR 3 R WR 4 R

 1   1 
 [ KH R ]   [ KH RC ] 
 WRH   WRH 

Solution: SYNC option (PARAM WR3, WR4 and WRH defined)

 
 [ BR ] + α1R [ M R ] + α 2R [ K R ] +   [ BRC ] + α1R [ M RC ] + α 2 R [ K RC ] + 
  
Ωβ R GR 
GR 1 1 1
iΩ  [KR ] + [ K 4R ] +   [ K RC ] + [ K 4CR ] + [ KH RC ] 
 WR3 WR4   WR3 WR 4 WRH 
 1   [ BRC ] + α1R [ M RC ] + α 2 R [ K RC ] + 
 [ KH R ]   
 WRH  +α R GR 1 1
 [ K RC ] + [ K 4CR ] + [ KH RC ] 
 WR3 WR 4 WRH 

Examples used in this chapter are available under the following folder: ../tpl/ugrotor/ and ../tpl/rotv18/. A
brief description of these examples is given below:

Table 7-10
Input File Name Description
rug_7_1 NM Rotor with CBEAM elements, Campbell Diagram (ROTORG).
rug_7_2 NM Rotor with CBAR elements, campbell Diagram (ROTORG).
rug_7_3 NM Rotor with CQUADX elements, campbell Diagram (ROTORAX).
rug_7_4 NM Rotor with Solid elements, campbell Diagram (ROTOR).
rug_7_5 NM Rotor with CBEAM elements, Synchronous Analysis (ROTORG).
rug_7_6 M Rotor with CBAR elements, Synchronous Analysis (ROTORG).
rug_7_7 NM Rotor with CQUADX elements, Synchronous Analysis (ROTORAX).
rug_7_8 NM Rotor with Solid elements,Synchronous Analysis (ROTOR).

Main Index
 Chapter 7: Complex Eigenvalue Analysis for Rotors 151
Summary for Equation of Motion

Table 7-10
Input File Name Description
rug_7_9 NM Rotor with CBEAM elements, Synchronous Analysis (ROTORG) - Case 2.
rug_7_10 NM Rotor with CBAR elements, (ROTORG)- Case 2.
rug_7_11 NM Rotor with CQUADX elements, (ROTORAX)- Case 2.
rug_7_12 NM Rotor with Solid elements, (ROTOR)- Case 2.
rug_7_13 NM Rotor with CBEAM elements, Synchronous Analysis (ROTORG) - Case 3,
Internal Damping.
rug_7_14 NM Rotor with CBEAM elements, Synchronous Analysis (ROTORG) - Case 3,
External Damping.
rug_7_15 CBEAM ROTOR in rotating co-ordinate system, Campbell Diagram.
rug_7_16 CBEAM ROTOR in rotating co-ordinate system, Forward whirl.
rug_7_17 CBEAM ROTOR in rotating coordinate system, Backward whirl.
rug_7_18 Shell model without differential stiffness.
rug_7_19 Shell model with differential stiffness.
../tpl/rotv18/r3dmd 3D NM rotor model for mode tracking
tr2.dat
../tpl/rotv18/r1d10 Mode Identification test case
7mdchk_30.dat

Main Index
 152 MSC Nastran Rotordynamics User’s Guide
Summary for Equation of Motion

Main Index
 Chapter 8: Frequency Response Analysis
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Frequency Response
8 Analysis


Introduction 154
 Equation used in the Analysis (Reference Frame) 154

Main Index
 154 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
Frequency response analysis is a method used to compute structural response to steady-state oscillatory
excitation. Frequency response analysis can be performed with the rotors spinning at user specified rates or
excitation that is synchronous with the reference rotor. In frequency response analysis the excitation is
explicitly defined in the frequency domain. All of the applied forces are known at each forcing frequency.
Forces can be in the form of applied forces or synchronous loads due to unbalance in rotor.
In MSC Nastran, direct frequency response can be obtained using SOL 108 while modal frequency response
analysis can be performed using SOL 111. Note that SOL 111 option is available for rotors defined using
ROTORG only.

Asynchronous Analysis
Calculation of response with the rotors spinning at set speeds is performed using the asynchronous option in
RGYRO. The user specifies the rotor speeds, and the corresponding gyroscopic terms are determined by the
code and added to the equation of motion, and the response is determined. Asynchronous analysis is required
when the excitation frequency is independent of the rotor speed and usually originating from outside the
rotor component.

Synchronous Analysis
Calculation of frequency response with the excitation synchronous with the reference rotor is performed
using the synchronous option in RGYRO. The user specifies the rotor speed range, the gyroscopic terms are
determined for each frequency and added to the equation of motion. There are two methods for determining
the gyroscopic terms. The first method calculates the terms for each frequency and adds them to the equation
of motion (default option). The second method estimates the gyroscopic terms using a least-mean-squared
fit of the terms within the analysis range (PARAM, GYROAVG < 0).

Equation used in the Analysis (Reference Frame)

Asynchronous Analysis
For frequency response with asynchronous excitation, Ω for each rotor is constant and can be determined
from the rotation speed of the reference rotor, Ωref, and relative rotation rates specified by the user. The
equation of motion to be solved is:

( −ω [M ] + iω (Ω(ω )[G ] + [ B ] + [ B ]) + ([ K ] + Ω(ω )[ K

2
S S ] ) ){g (ω )} = F (ω )
C R

ω Natural frequency of the system

[M] Mass matrix
i
Ω(ω)Rotor Speed

Main Index
 Chapter 8: Frequency Response Analysis 155
Equation used in the Analysis (Reference Frame)

[G] Gyroscopic matrix

[BS] Stator damping matrix
[BR] Rotor damping matrix
[K] Stiffness matrix
[KC]R Circulation terms due to damping in rotating system
g(ω)Complex frequency response
Expanding the above equation to include all options available to define damping

 −ω 2 [ M ] + 
 
  ([ BS ] + α1[ M S ] + α 2 [ K S ] + [ BH S ]) +  
 iω  n  
  +
   (
j =1
)
Ω j (Ω ref ) [G ] j + [ BR ] j + [ BH R ] j + α1Rj [ M R ] + α 2 Rj [ K R ] j  
 

 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] +  {g (ω )} = { F (ω )}
 
  [ K R ] j (1 + iGR j ) + i[ K 4 R ] j + i[ KH R ] j  
 n   
  BRC  +  BH RC  + α1Rj  M RC  + α 2 Rj  K RC  +  
           
 j =1  +Ω j (Ω ref ) 
j j j

 
   GR 1 1 
   K R  +  K 4 R  +  KH R 
C C C
 
   ω j ω j ω j  
where j references individual rotors. The factor 1/ω will be determined for each excitation frequency, similar
to frequency-dependent elements.
For the option to bypass the frequency-dependent lookup of rotor speeds (PARAM,GYROAVG,-1), the
equation of motion to be solved is:

 −ω 2 [ M ] + 
 
  ([ BS ] + α1[ M S ] + α 2 [ K S ] + [ BH S ]) +  
   
 iω  n  Ω j (Ω ref ) [G ] j + [ BR ] j + [ BH R ] j + α1Rj [ M R ] + α 2 Rj [ K R ] j +   + 
    
   GR
[K ] +
1
[ K 4R ] j +
1
[ KH R ] j  
   WR3 R j WR 4    {g (ω )} = F (ω )
j =1
WRH  { }
 
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] + 
    BR  +  BH R  + α1Rj  M R  + α 2 Rj  K R  +   
C C C C
 n   j  j     j 
   [ K R ] j + Ω j (Ω ref )  
 j =1  GR j 1 1 
 K R  j + WR 4  K 4 R  j + WRH  KH R  j   
C C C
 
   WR 3 

Main Index
 156 MSC Nastran Rotordynamics User’s Guide
Equation used in the Analysis (Reference Frame)

The main difference here is the usage of parameters WR3, WR4 and WRH for structural damping terms.
Use of these parameters avoids the division by excitation frequency observed in the previous equation. Also,
in case the parameters WR3, WR4 and WRH are not defined in the analysis, then the corresponding
structural damping terms are not included in the analysis.

Synchronous Analysis
For frequency response with synchronous excitation, the excitation frequency is equal to the spin rate of the
reference rotor, that is, ω = Ωref. The spin rates of the additional rotors can be determined from the relative
spin rates specified by the user. The equation of motion to be solved is:

( −ω [M ] + iω ([B ] + [ B ] + Ω [G]) + ([K ] + Ω[ K

2
S R j ]
C R ) ){g (ω )} = {F (ω )}
Substituting ω = Ω

( −Ω ref
2
)
[ M ] + iΩ ref ([ BS ] + [ BR ] + Ω j (Ω ref )[G ]) + ([ K ] + Ω j (Ω ref )[ K C ]R ) {g (Ω ref )} = { F (Ω ref )}

The above equation can be expanded based on different options available for defining damping for the stator
and for rotors.

 −ω 2 [ M ] + 
 
  ([ BS ] + α 1[ M S ] + α 2 [ K S ] + [ BH S ]) +  
 iω  n  
  +
   j =1
( )
Ω j (ω ) [G ] j + [ BR ] j + [ BH R ] j + α 1Rj [ M R ] + α 2 Rj [ K R ] j  
 

 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] +  { g (ω )} = { F (ω )}
 
  [ K R ] j (1 + iGR j ) + i[ K 4 R ] j + i[ KH R ] j  
 n    BRC  +  BH RC  + α 1Rj  M RC  + α 2 Rj  K RC  +  
 
  
 j  j    j 
 j =1  +Ω j (ω )  
 
   GR 1 1 
   K R  +  K 4 R  +  KH R 
C C C
 
   ω  j ω  j ω j  

As compared to the asynchronous analysis, the gyroscopic and the circulation terms for synchronous analysis
are treated differently.
For the option to bypass the frequency-dependent lookup of rotor speeds (PARAM,GYROAVG,-1), Ωj for
each rotor is written as a linear function dependent on the reference rotor spin rate:

Ω j = α j + β j Ωref
The scaling factor, a j and βj, is determined from the relative spin rates specified by the user on the RSPINR
entries. The Ωref in the 1/ Ωref terms are replaced by the values of user parameters WR3 and WR4. The
equation of motion to be solved is:

Main Index
 Chapter 8: Frequency Response Analysis 157
Equation used in the Analysis (Reference Frame)

 2
n  
 −Ω ref  [ M ] − i  β j [G ] j  + 
  j =1  
 
 

([ S ] [ S ] [ S ] [ S ])
B + α 1 M + α 2 K + BH + 


 
    [ BR ] j + [ BH R ] j + α 1Rj [ M R ] j + α 2 Rj [ K R ] j +    
       
   α j [G ] j +  GR j 1 1  + 
 iΩ ref  n   [ K ] + [ K 4 ] + [ KH ]
R j  
  +
R j R j
   WR3 WR 4 WRH
 
 j =1    BRC  +  BH RC  + α 1Rj  M RC  + α 2 Rj  K RC  +     g a (Ω ref ) = 0

        
    j j j j   
   −i β j  GR 1 1   
   K R  +  K 4 R  +  KH R     
j C C C
  
   WR3 j WR 4 j WRH j
 
 
 (1 + iG )[ K S ] + i[ K 4 S ] + i[ KH S ] + 
 
 n    BRC  +  BH RC  + α 1Rj  M RC  + α 2 Rj  K RC  +  
 j  j  j   j  
 [K ] + α  
  R j j  GR 1 1  
 j =1 
  K R  +
C
 K 4 R  +
C
 KH R   
C

   WR3 j WR 4 j WRH j  

As in the asynchronous case with GYROAVG parameter, in case the parameters WR3, WR4 and WRH are
not defined in the analysis, then the corresponding structural damping terms are not included in the analysis.

Example: NM (Nelson-McVaugh) Rotor

In this example, frequency response analysis is performed to obtain displacement in response to synchronous
excitation by the unbalance loads for the NM rotor described earlier. Here are the changes made in the
analysis deck to define synchronous unbalance loads
Case Control
FREQ = 100
DLOAD = 10
Bulk Data
FREQ1 100 0. 1.0 1000
RGYRO 1 SYNC 4 RPM 10000.0
$ UNBALANCE LOADS
UNBALNC 10 122.68 41 1. 0. 0.
1.-4 0.0 0.0 0.0 1000.
As described in the previous chapter, for synchronous frequency response analysis, the DLOAD in case
control points to an unbalance entry in the bulk data. Also, the GYRO field needs to show “SYNC” as the
analysis option.
For this particular case, the variation of unbalance loads and displacement at the point of unbalance with
frequency is shown in Figure 8-1. It can be seen here that the load due to unbalance increases with the square
of the excitation frequency. Also, the displacement results show a peak at the first forward whirl critical speed,
as expected.

Main Index
 158 MSC Nastran Rotordynamics User’s Guide
Equation used in the Analysis (Reference Frame)

Figure 8-1 Loads due to unbalance and corresponding displacement

Effect of Damping
In the next step, damping is introduced in the analysis through the spring supports. It is possible to define
damping in the fixed or rotating reference frame though CBUSH elements as shown in Figure 8-2. If the
damping is defined in the rotating coordinate reference frame, then the corresponding circulation terms are
included for analysis in the equations of motion.

Figure 8-2 Defining spring supports in fixed and rotating reference frame

Main Index
 Chapter 8: Frequency Response Analysis 159
Equation used in the Analysis (Reference Frame)

Results obtained with external and internal damping are shown in Figure 8-3. As compared to results obtained
in Figure 8-1 for displacement, the displacement obtained in Figure 8-3 is smaller. For this particular case, the
difference between the displacements obtained with external and internal damping is small.

Figure 8-3 Effect of internal and external damping

Effect of GYROAVG parameter

The case with external damping was analyzed again with “PARAM, GYROAVG, -1”. The results obtained
for this case show small difference in results, see Figure 8-4. However, the computational time required for this
problem drops from 0.08 sec to 0.01 sec. Thus, this feature is useful for obtaining quick approximate results.

Figure 8-4 Effect of “Param, GYROAVG, -1”

Main Index
 160 MSC Nastran Rotordynamics User’s Guide
Equation used in the Analysis (Reference Frame)

Frequency Dependent Bearings

For linear frequency response analysis, linear frequency dependent bearing properties can be defined using
the PBUSHT card. Brief description of the PBUSHT card, which defines properties for the CBUSH elements
is shown in Figure 8-5.

PBUSHT Frequency Dependent or Nonlinear Force Deflection Spring and Damper Property

Defines the frequency dependent properties or the stress dependent properties for a generalized spring and
damper structural element.

Format:
1 2 3 4 5 6 7 8 9 10
PBUSHT PID “K” TKID1 TKID2 TKID3 TKID4 TKID5 TKID6
“B” TBID1 TBID2 TBID3 TBID4 TBID5 TBID6
“GE” TGEID1 TGEID2 TGEID3 TGEID4 TGEID5 TGEID6
“KN” TKNID1 TKIND2 TKNID3 TKIND4 TKIND5 TKIND6
FDC FUSE DIR OPTION LOWER UPPER
FSRS LRGR

Figure 8-5 PBUSHT entry to define frequency dependent bearing properties

Modifications made in the bulk data to introduce the PBUSHT entry are shown below:
CBUSH 201 211 1091 0
CBUSH 202 211 1111 0

PBUSHT 211 K 1001 1001

B 1003 1003
PBUSH 211 K 4.38+7 4.38+7
B 2627.0 2627.0
TABLED1 1001
0. 4.38+7 250. 4.38+7 1000. 8.38+7 ENDT
TABLED1 1003
0. 2627.0 250. 2627.0 1000. 1627.0 ENDT
Note that both PBUSH and PBUSHT entries are defined in the analysis. For this analysis, the bearing
properties are kept unchanged till 250 Hz. Thereafter, the bearing stiffness has been increased while the
damping has been slightly reduced. Effect of these changes can be seen in Figure 8-6 where the resonance peak
has shifted to the right due to increase in stiffness. Also there is small increase in peak amplitude due to
reduction in bearing damping.

Main Index
 Chapter 8: Frequency Response Analysis 161
Equation used in the Analysis (Reference Frame)

Figure 8-6 Use of PBUSHT entry for frequency dependent bearing properties

Modal Analysis (for ROTORG only)

If the model includes only line rotors defined using ROTORG, then modal frequency response analysis can
also be performed using SOL 111. In order to transform a test deck from direct frequency response analysis
to SOL 111 test deck, the following changes are required:

In Case Control
Change the solution sequence number and include a method card
SOL 111
CEND
METHOD = 100

Bulk Data
Include real eigenvalue analysis in the deck which points to METHOD defined in the case control.
EIGRL 100 10

Main Index
 162 MSC Nastran Rotordynamics User’s Guide
Equation used in the Analysis (Reference Frame)

The results obtained using SOL 111 show forces and displacement requested as part of output, just like direct
frequency response analysis. It also shows real eigenvalue analysis performed to extract the mode shapes for
modal analysis (see Figure 8-7).

Figure 8-7 Real Eigenvalue analysis performed for extracting modeshapes

Modal frequency response was performed using 10 eigenvalues and again with 50 eigenvalues. Results
obtained for these cases are shown in Figure 8-8 and compared with those obtained using SOL 108. It can be
seen that for all the three cases, the response is almost identical.

Main Index
 Chapter 8: Frequency Response Analysis 163
Equation used in the Analysis (Reference Frame)

Figure 8-8 Displacement obtained from real and modal frequency response analysis

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 8-1
Input File Name Description
rug_8_1 Unbalance response for 1D beam rotor.
rug_8_2 Unbalance response for 1D beam rotor with external damping.
rug_8_3 Unbalance response for 1D beam rotor with internal damping.
rug_8_4 Unbalance response for 1D beam rotor with external damping and param
GYROAVG.
rug_8_5 Use of PBUSHT.
rug_8_6 Modal Analysis

Main Index
 164 MSC Nastran Rotordynamics User’s Guide
Equation used in the Analysis (Reference Frame)

Main Index
 Chapter 9: Transient Response Analysis
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

9 Transient Response Analysis


Introduction 166

Equation of motion 166
 Input file for transient analysis 169

Comparison with axisymmetric and 3D rotors 174

Usage of Nonlinear Elements 175

Main Index
 166 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
Transient analysis with rotordynamics includes gyroscopic terms and option to specify rotor imbalance. Both
nonlinear and direct linear transient response solutions internally calculate the gyroscopic and unbalance
force terms for each time step. The transient excitation is explicitly defined in the time domain. All of the
external forces applied to the structure are known at each instant in time. Transient analysis also supports
usage of nonlinear bearing elements like NLRGAP, NLRSFD and NOLINi. Details related to general
transient analysis in MSC Nastran in provided in MSC Nastran Dynamic Analysis User’s Guide . Only details related
to rotordynamic analysis are provided in this section. Note that the modal transient solution sequence, SOL
112, is not supported for rotordynamic analysis. However, external SE approach can be used to obtain model
reduction, if required.
For transient analysis with rotordynamics, use of nonlinear transient solver like SOL 400 is recommended
due to enhanced numerical stability.

Equation of motion
Equation of motion solved in transient analysis is given by:
[ M ]{ g(t )} + ([ BS ] + [ BR ] + Ω (t )[G ]){ g (t )} + ([ K ] + Ω (t )[ K C ]R + Ω
 (t )[ K T ]) { g (t )} = {F (t )}
S

where, [KT] is a skew symmetric matrix and for each grid point in the model it is defined by:

0 0 0 0 0 0
0 0 0 0 0 0 

0 0 0 0 0 0
 K rT  =  
0 0 0 0 0 0
0 0 0 0 0 I11 
 
0 0 0 0 − I11 0 

Main Index
 Chapter 9: Transient Response Analysis 167
Equation of motion

Note that the equation of motion used for the transient analysis includes effect of rotor spin rate. Expanded
version of the above equation is given by:
[M ]{g(t )} +
 G 1 1 
 ([ BS ] + α1[ M S ] + α 2 [ KS ] + [ BH S ]) + W 3 [ KS ] + W 4 [ K 4S ] + WH [ KH S ] 
 
 n  Ω j (t ) [G] j + [ BR ] j + [ BH R ] j + α1Rj [ M R ] + α 2Rj [ KR ] j +  {g (t )} +
   
  GR 1 1  
 j =1  [ KR ] j + [ K 4R ] j + [ KH R ] j  
  WR3 WR4 WRH  
 [ KS ] + 
 
 n    BRC  +  BH RC  + α1Rj  M RC  + α 2Rj  KRC  + 
 j  j    j 

  [ K ] + Ω (t )   (t )[ K T ]  {g (t )} = {F (t )}

  R j j  GR j 1 1  +Ω j R j 

 j =1    KRC  +
  K 4CR  +
  KH RC  
 
   WR 3 j WR 4 j WRH j 


where,
[M] Total mass matrix for the model
[BS] Viscous damping from dampers in the stator portion
[MS] Mass matrix for the stator portion
[KS] Stiffness matrix for the stator portion
[BHS] Hybrid damping for the stator portion
α1, α2 Rayleigh damping parameters for stator portion
αj(t) Rotor speed for the jth rotor
[G] Gyroscopic matrix for the rotor
[BR] Viscous damping from dampers for the rotor
[MR] Mass matrix for the rotor
[KR] Stiffness matrix for rotor
[BHR] Hybrid damping for the rotor
α1R, α2R Rayleigh damping parameters for the rotor
GR Structural damping for the rotor defined in RSPINR
[K4R] Material damping for the rotor part
[KHR] Hybrid damping for the rotor part
WR3, WR4, WRH Parameters to convert structural damping to equivalent viscous damping.
G Structural damping for the stator portion
[K4S] Material damping for the stator portion
[KHS] Hybrid structural damping for stator
Circulation matrix due to [BR]
 BRC 
Circulation matrix due to [BHR]
 B H RC 
Circulation matrix due to Rayleigh damping for rotors
α1[ M RC ] + α 2 [ K RC ]
GR
 K RC 
WR 3  Circulation matrix due to structural damping for rotors

Main Index
 168 MSC Nastran Rotordynamics User’s Guide
Equation of motion

1
 K 4CR 
WR4 
Circulation matrix due to structural damping for rotors

1
 KH RC 
WRH 
Circulation matrix due to hybrid damping for rotor
{g (t)} State vector for transient analysis (D-set DOF)
F(t) Linear Loads and Nonlinear forces from bearings and connectors

Unbalance Loads for Transient Analysis

Unlike its use in the frequency response, complex modes, and static solutions, the RGYRO (Case) does not
reference an RGYRO Bulk Data entry. Instead, the RGYRO Case Control command references an UNBALNC
Bulk Data entry. A brief description of the unbalance entry from QRG is shown in Figure 9-1.

UNBALNC Specifies an Unbalanced Load for Rotordynamic Transient or Frequency Response

Analysis

Used in rotordynamic analyses to specify a rotating unbalance load in terms of a cylindrical system with the
rotor rotation as the z-axis.

Format:
1 2 3 4 5 6 7 8 9 10
UNBALN SID MASS GRID X1 X2 X3
C
ROFFSET THETA ZOFFSET T/Fon T/FOFF
UFT1 UFT2 UFT3 UFR1 UFR2 UFR3
MCT1 MCT2 MCT3 MCR1 MCR2 MCR3
SCR1 SCR2 SCR3

Figure 9-1 Unbalance (UNBALNC) Entry for transient analysis

Variables relevant to describe the unbalance loads are described below:

MASS Mass of imbalance (m)
GRID Grid identification number for applying imbalance
X1, X2, X3 Components of vector used to define cylindrical coordinate system centered at GRID
ROFFSET Offset mass in the radial direction (r) in the unbalance coordinate system
ZOFFSET Offset mass in the Z direction (Z) in the unbalance coordinate system
THETA Angular position, in degrees, of the mass in the unbalance coordinate system.(θ)
UFT/UFR EPOINT to output unbalance forces and moments (for transient analysis only)

Main Index
 Chapter 9: Transient Response Analysis 169
Input file for transient analysis

Using these variables, forces and moments generated by unbalance loads can be expressed as:

 F1 (t )  m(t )r (t )(Ω 2j (t ) cos(θ jk (t )) + Ω  (t ) sin(θ (t ))) 

j j
  = WTMASS   
 m(t )r (t )(Ω j (t ) sin(θ jk (t )) − Ω j (t ) cos(θ j (t ))) 
2
 F2 (t ) 
 M 1 (t )  0 −1  F1 (t ) 
 =Z  
 M 2 (t )  1 0   F2 (t ) 

where, θ j ( t ) =  Ω j ( t ) dt + θ (angular position for kth unbalance load)

0

Input file for transient analysis

The NM rotor described earlier is used again for the transient analysis. In this case, external loading is
obtained through the rotor imbalance, but any generic time dependent load can also be used in the analysis.

Figure 9-2 NM Rotor model for transient analysis

Changes required in different sections of the input file for transient analysis are described below:
a. Case Control Section
ECHO = SORT
SPC = 2
RGYRO = 10
TSTEP = 1
SET 99 = 41
SET 98 = 95,92,93,94
DISP(SORT2,PHASE)=99
OLOAD(SORT2,PHASE)=99
NLLOAD(SORT2,PHASE)=99
SDISP(SORT2) = 98

Main Index
 170 MSC Nastran Rotordynamics User’s Guide
Input file for transient analysis

b. Time Step and rotor speed definition in Bulk Data

TSTEPNL 1 100000 .00001 10
+
+ 0
$ Rotors Relative Spin Rates
RSPINT 4 9 10 FREQ 1 95
TABLED1 1
0.0 0.0 0.005 0.0 .5 400. 1000. 400.
ENDT

c. Unbalance Load definition

$ UNBALANCE LOADS
UNBALNC 10 122.68 41 1. 0. 0.
1.-4 0.0 0.0 0.0 1000.
92 93 94
EPOINT 92 93 94 95
Results obtained from the transient analysis are described below. Here also, we have considered the effect of
both internal and external damping, as described in the previous chapter. (also presented in Figure 9-3)

Figure 9-3 External and internal Damping for rotordynamic analysis

Main Index
 Chapter 9: Transient Response Analysis 171
Input file for transient analysis

Effect of External Damping

Variation of unbalance load and displacement with time is shown in Figure 9-4. The spin rate defined earlier
increases from 0 to 400 Hz in 0.5 seconds and then it stays constant. The unbalanced loads observed follow
the expected trend. The loads increase with square of rotor speed till 0.5 secs and then they stay constant for
rest of the analysis. Further insight into the results is presented in Figure 9-5 to Figure 9-8.

Figure 9-4 Unbalance loads and corresponding displacement from transient analysis

Figure 9-5 X-Y components of unbalance load

Figure 9-5 shows the variation of loads in X-Y direction due to rotor unbalance. Note that rotor is oriented
along the Z-direction and XY plane is perpendicular to the axis of rotation. It can be seen in the results that
loads in X and Y direction have a phase difference of 90 deg, as expected. The loads in the Y direction are
lagging behind by 90 deg.

Main Index
 172 MSC Nastran Rotordynamics User’s Guide
Input file for transient analysis

Figure 9-6 Displacement near the critical speed region

The displacement shown in Figure 9-4 show increase in amplitude of displacement with time as the rotor speed
increases. However, after 0.37 seconds, there is decrease in amplitude and the amplitude is steady after 0.5
seconds. The variation of displacement between 0.35 to 0.55 seconds is shown in Figure 9-6. It highlights the
transition zone described earlier. This behavior for displacement can be explained by plotting displacement
along with rotor speed, as shown in Figure 9-7 and Figure 9-8. In these figures, the right axis (in green) shows
variation of rotor speed provided as input. The results presented in these figure show that the peak amplitude
in displacement is observed near 0.375 sec where the rotor speed is around 285 Hz. This frequency coincides
with the first forward whirl critical speed observed in the complex eigenvalue analysis of the rotor model.

Figure 9-7 Variation of rotor speed and displacement with time

Main Index
 Chapter 9: Transient Response Analysis 173
Input file for transient analysis

Figure 9-8 Variation of rotor speed and displacement with time near the forward whirl critical speed

Effect of Internal Damping

Effect of internal damping is studied by connecting bearing to the rotor directly in the rotating reference
frame (hypothetical case). Here two different cases are considered. In case 2, the damping and stiffness due
to bearings is same as in case 1, but the effect of circulation terms is reduced by half.
a. Internal damping 1
CBUSH 201 211 91
CBUSH 202 211 111
PBUSH 211 K 4.38+7 4.38+7
B 2627.0 2627.0
b. Internal Damping 2
CBUSH 201 211 91
CBUSH 202 211 111
CBUSH 203 211 1091
CBUSH 204 211 1111
PBUSH 211 K 2.19+7 2.19+7
B 1313.5 1313.5
Results for these two cases with internal damping are shown in Figure 9-9 and compared with those obtained
using external damping (which does not produce any circulation matrix). Here it can be seen that for case 1,
displacements diverge after the first forward whirl critical speed. However, for case 2, where the effect of
circulation terms is reduced by half, the results do not diverge. But they show higher vibration as compared

Main Index
 174 MSC Nastran Rotordynamics User’s Guide
Comparison with axisymmetric and 3D rotors

to the externally damped bearing case. Thus, internal damping needs to be carefully analyzed as it may lead
to higher vibration or instability in the rotordynamic analysis.

Figure 9-9 Effect of internal and external damping

Comparison with axisymmetric and 3D rotors

The NM rotor used in this analysis can also be modeled using 2D axisymmetric harmonic and solid elements
as described earlier. Response obtained using these higher order elements is shown in Figure 9-10. It can be
seen here that the maximum amplitude predicted by these three analyses are close to each other.

Main Index
 Chapter 9: Transient Response Analysis 175
Usage of Nonlinear Elements

Figure 9-10 Transient Response obtained using higher order elements

Usage of Nonlinear Elements

In transient analysis, it is possible to include nonlinear bearing elements for rotordynamic analysis. Nonlinear
bearing elements can be specified using NLRGAP, NLRSFD and NOLINi (p. 2154) entries.
In the previous example, the maximum displacement amplitude obtained is around 0.08 units. As a
hypothetical case, consider a scenario where the maximum amplitude is restricted to 0.05 units using a very
stiff spring. A behavior like this can be modelled using the nonlinear NLRGAP element. For simplicity, beam
rotor model defined using ROTORG entry is used in the analysis. Changes required in the input file to
include the above mentioned nonlinear NLRGAP element is described below:

Figure 9-11 Introducing nonlinear bearing element in the analysis

Changes in case Control

NONLINEAR = 1001

Changes in Bulk data entries

$ NLRGAP
GRID 141 0. 0. 0.0889

Main Index
 176 MSC Nastran Rotordynamics User’s Guide
Usage of Nonlinear Elements

GRID 1141 0. 0. 0.0889

RBE2 303 141 123456 41
NLRGAP 1001 141 1141 XY 11 12
CBUSH 203 212 1141 0
PBUSH 212 K 1.0+20 1.0+20
TABLED1 11
0.0 1.0+12 1000. 1.0+12 ENDT
TABLED1 12
0.0 0.05 1000. 0.05 ENDT
Displacement obtained with the nonlinear NLRGAP element is shown in Figure 9-12 and Figure 9-13. Results
show that, although the displacement of Grid 41 is reduced to 0.05, it does not decay after crossing the first
forward whirl critical speed as observed in the earlier case.

Figure 9-12 Effect on nonlinear NLRGAP element on transient response

Main Index
 Chapter 9: Transient Response Analysis 177
Mass and speed correction for Unbalance Loads

Figure 9-13 Displacement obtained using nonlinear NLRGAP element

Mass and speed correction for Unbalance Loads

The UNBALNC entry with CFLAG option and EPOINTS to output the correction terms is shown
below.
UNBALNC 10 122.68 41 1. 0. 0.
1.-4 0.0 0.001 0.0 1000. CFLAG
202 203 204 205 206 207
208 209 210 211 212 213
214 215 216
EPOINT 201 THRU 216

Here, the CFLAG field can take following values: NONE (Default), SPEED, MASS and BOTH. In this
particular case, the value of mass used on the UNBALNC entry is unusually high while the radial distance
used is small to highlight how the correction terms can influence the results. Without the correction terms,
the unbalance forces generated depend only on the product of unbalance mass with radial distance. However,
when correction terms are included, both mass and variation of rotor speed with time can influence the
unbalance loads generated.
The unbalance loads generated (as output through various EPOINTS) are shown in Figure 9-14 to Figure 9-18.
As expected, the basic unbalance loads generated without the correction terms are identical in all the four
cases considered, see Figure 9-14 and Figure 9-15. The effect of mass correction on forces and moments is shown
in Figure 9-16 and Figure 9-17, respectively. Here, the results obtained using CFLAG=MASS and
CFLAG=BOTH show large forces and non-zero moments being generated due to mass correction terms.
Similarly, CFLAG=NONE and CFLAG=SPEED produce similar results since they do not include mass

Main Index
 178 MSC Nastran Rotordynamics User’s Guide
Mass and speed correction for Unbalance Loads

corrections. The moments produced due to speed correction terms is shown in Figure 9-19. It can be seen here
that only CFLAG=SPEED produces noticeable moments due to speed correction. The moments generated
in this case are dependent on the variation of rotor speed with time and rotational displacement. As a result,
the moment produced diminishes to zero after 0.5 seconds after which the rotor speed stays constant. The
case with CFLAG=BOTH also generates moments due to speed correction, however due to smaller rotational
displacements (see Figure 9-19 and Figure 9-20), its magnitude is negligible as compared to the case where
CFLAG=SPEED. The displacement and rotation due to unbalance loads for different CLFAG options are
shown in Figure 9-19 and Figure 9-20, where the effect of large unbalance mass can be observed for the case
where CFLAG=MASS and CFLAG=BOTH.

Figure 9-14 Variation of unbalance forces without correction (Output from EPOINTs 202, 203 & 204)

Main Index
 Chapter 9: Transient Response Analysis 179
Mass and speed correction for Unbalance Loads

Figure 9-15 Variation of unbalance moments without correction (Output from EPOINTs 205, 206 & 207)

Figure 9-16 Forces generated due to mass correction (Output from EPOINTs 208, 209 & 210)

Main Index
 180 MSC Nastran Rotordynamics User’s Guide
Mass and speed correction for Unbalance Loads

Figure 9-17 Moments generated due to mass correction (Output from EPOINTs 211, 212 & 213)

Figure 9-18 Moments generated due to speed correction (Output from EPOINTs 214, 215 & 216)

Main Index
 Chapter 9: Transient Response Analysis 181
Mass and speed correction for Unbalance Loads

Figure 9-19 X-Displacement due to different CFLAG options

Figure 9-20 X-Rotation due to different CFLAG options

Main Index
 182 MSC Nastran Rotordynamics User’s Guide
Mass and speed correction for Unbalance Loads

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 9-1
Input File Name Description
rug_9_1 Analysis with unbalance loads in SOL 129.
rug_9_2 Analysis with unbalance loads in SOL 400.
rug_9_3 Analysis with unbalance loads in SOL 129 with internal Damping - case 1.
rug_9_4 Analysis with unbalance loads in SOL 129 with internal Damping - case 2.
rug_9_5 Unbalance analysis with ROTORAX.
rug_9_6 Unbalance analysis with solid rotor elements.
rug_9_7 Transient analysis with nonlinear bearing element.

Main Index
 Chapter 10: Nonlinear Frequency Response
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Nonlinear Frequency
10 Response


Introduction 184
 User Interface 185

Analysis Setup 189

Theoretical Details 193
 Examples 195

Main Index
 184 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
There is a class of dynamic response analyses where a structure exhibiting nonlinearities is subjected to a
harmonic excitation in which the response is essentially periodic. That is to say that the degree of nonlinearity
in the system is light enough that the response may be described sufficiently accurately by a combination of
harmonic responses, i.e. periodic; this is achieved using Fourier series.
Some examples of this class of dynamic response problems are rotor/stator contact under abnormal running
conditions, an overload condition in an oscillating mechanism causing periodic contact or perhaps just
simply a nonlinear bearing stiffness. The response of such systems may exhibit multiple solutions in a steady-
state vibration response scenario, possibly with amplitude jumps as the system moves from one frequency to
another, such as might occur in a rotor that is increasing or decreasing in speed. These jumps reveal different
behavior of the dynamic system with increasing or decreasing excitation frequency.
The nonlinear harmonic response solution sequence uses the harmonic balance method to calculate the
periodic response of a non-linear system under harmonic excitation. This requires the definition of a
frequency domain problem in the presence of nonlinearities. The harmonic balance method assumes the
steady-state response consists of a sum of sinusoidal responses finding the coefficients of the sinusoids to
satisfy the equations of motion. Harmonic balance is only efficient if a small number of sinusoids are
necessary to approximate the solution to a desired accuracy and this is why the nonlinearities in the system
must be mild.
As with any nonlinearity, it must only be present in the residual structure, but this does not preclude the use
of superelements or ASET degrees of freedom to perform static condensation or dynamic reduction using
CMS (Component Mode Synthesis). In fact this is a recommended technique in order to keep the number
of degrees of freedom for harmonic balance to a minimum.
Nonlinear harmonic response is available in the presence or absence of rotors, but there must be at least one
degree of freedom defined on nonlinear force type entries such as the NLRGAP, NLRSFD and NOLINi
entries. Other types of nonlinearity that may be defined include the CBUSH2D element or indeed any
elements having frequency dependent properties.
MSC Nastran, which incorporates SCA technology, adds the possibility of creating a user defined service
whereby the NLRSFD entry is able to call external user defined behavior to replace the standard NLRSFD
behavior.
Its limitation therefore is that it can only capture harmonic components, so any frequencies which are not
pure sub- or super-harmonics of the excitation will be lost. It is also possible that either no solution exists (an
unstable dynamic system), or that more than one solution is possible. The nonlinear system being studied
may exhibit bifurcation or turning points, points for which a small incremental change in frequency results
in more than one solution. In addition the system may exhibit a step change in the response from one
frequency to the next, what is referred to in the literature as “jump phenomena”. In its present form, the
presence of bifurcation or turning points cannot be determined in nonlinear harmonic response analysis and
the resulting solution is just one of the possible states. Jump phenomena are revealed by examining the
response curves.
In a system with bearing clearance, like hydrodynamic or magnetic bearings, the stiffness of the system
changes depending on whether the bearing forces overcome the clearance in the bearing resulting in contact
or no-contact conditions. Each of the contact/no-contact states may have essentially linear behaviour, but the

Main Index
 Chapter 10: Nonlinear Frequency Response 185
User Interface

overall behavior is nonlinear if the system changes from one state to another. Probably the simplest example
of such a system is the Duffing oscillator. The Duffing equation is:
3
ax·· + bx· + cx + dx = f cos ωt (10-1)
and is defined in any text book on nonlinear dynamics. This equation exhibits jump phenomena for certain
frequency values f where the solution “jumps” from one significantly different value to another for a small
change in frequency. To complicate matters, the jump differs depending on whether the change in frequency
is positive or negative.
When the system is nonlinear, there is also the possibility of sub and super harmonic responses. The response
frequencies in linear harmonic analysis are the same as the forcing frequency. Permanent oscillations whose
frequencies are a fraction of the forcing frequency (½, ¼, ...) may occur in a nonlinear system; these
oscillations are known as sub-harmonic response. Sub-harmonic responses require special conditions (e.g.
particular damping characteristics and a non-symmetric stiffness) whereas super-harmonic responses, i.e.
permanent oscillations whose frequencies are a multiple of the forcing frequency (2, 3, …), are always present;
whether they show significant response levels depends on the damping in the system. These are complex
phenomena which may in themselves exhibit jumps further complicating matters. Refer to book on Shock
and Vibration (Harris, 1987) for a more complete discussion of sub and super harmonic responses.
Nonlinear harmonic response uses an iterative procedure to find the coefficients for the combination of
sinusoids that form the steady-state response. Newton’s method of iteration is employed to solve a system of
nonlinear algebraic equations. A trial solution is attempted for displacements and the corresponding forces
in any NLRGAP, NLRSFD or NOLINi entries are calculated from the user supplied data. The residual forces
in the system are calculated and a convergence error is obtained. Based on the size of the convergence error,
the solution is either accepted or an updated displacement scaling is calculated and the new displacements
calculated. The sequence loops until either a diverging system is detected or convergence obtained.

User Interface
Frequency dependent loading for SOL 128 can be applied through standard forcing functions available in
MSC Nastran or through the unbalance loads in case rotors are present in the model.

File Management Section (FMS)

One modified FMS entry, related to the SCA UDS (Service Component Architecture User Defined
Subroutine), is required if a user defined service is to be used. When a user defined service (UDS) is to be
used to describe the behavior of the NLRSFD type nonlinearity in nonlinear harmonic response, the FMS
CONNECT entry is used with the SERVICE qualifier to make the connection between the GROUP name
on the NLRSFD bulk data entry and the name of the external service.

Executive Section
Nonlinear harmonic response is available as SOL 128 or SOL SENLHARM. This is a complete solution
sequence based around linear frequency response (SOL 108) in which nonlinearities may be taken into
account.

Main Index
 186 MSC Nastran Rotordynamics User’s Guide
User Interface

Case Control
The case control entry NLHARM has been added for nonlinear harmonic response to reference the
NLHARM bulk data entry. The existing NONLINEAR case control entry, previously dedicated to transient
response analysis, may now also be used in nonlinear harmonic response to reference nonlinear force bulk
data entries (NOLIN1, NOLIN2, NOLIN3, NOLIN4, NLRGAP, NLRSFD).
NLHARM
The NLHARM case control entry is dedicated to nonlinear harmonic response; it is used to reference a bulk
data NLHARM entry.
NLHARM=n
where n is the ID of the NLHARM bulk data entry.
NONLINEAR
The existing NONLINEAR case control entry, which was previously dedicated to transient linear and
transient nonlinear analysis, may now be used in nonlinear harmonic response analysis.
The NONLINEAR case control command is used in exactly the same way in nonlinear harmonic response
as it is in transient response to refer to the collection of nonlinear force entries (NLRGAP, NLRSFD and
NONLINi).

Bulk Data Entries

There are three bulk data entries (NLHARM, NLFREQ and NLFREQ1) relating specifically to nonlinear
harmonic response. The NOLINi bulk data entries may now be used in nonlinear harmonic response, but
they do not require any special remarks.

NLHARM
The NLHARM bulk data entry is used to define the parameters for nonlinear harmonic response.
1 2 3 4 5 6 7 8 9 10
NLHARM ID SUBFAC NHARM NLFREQ
ID Identification number referenced by the NLHARM case control entry (integer > 0).
SUBFAC Factor for capturing sub-harmonic response (Integer >=1, Default= 1). See remark 3.
NHARM The number of harmonics to include in the solution (integer > 0). See remark 2.
NLFREQ Identification number of the NLFREQ or NLFREQ1 entry specifying the forcing
frequency list (integer >0).

Remarks
1. The NLHARM ID must be unique among all NLHARM entries.

Main Index
 Chapter 10: Nonlinear Frequency Response 187
User Interface

2. The greater the degree of nonlinearity, the larger the number of harmonics required to find a solution.
The number of harmonics (NHARM) may therefore influence the ability of the algorithm to find
solutions to the nonlinear problem posed. However, the larger the number of harmonics, the more
computational effort is required to compute a solution.
3. The response frequencies in linear harmonic analysis are the same as the forcing frequency. Permanent
oscillations whose frequencies are a fraction of the forcing frequency (½, ¼, ...) may occur in a
nonlinear system; these oscillations are known as sub-harmonic response (Harris, 1987).

NLFREQ
The NLFREQ entry is used to define the forcing frequencies for nonlinear harmonic response.
1 2 3 4 5 6 7 8 9 10
NLFREQ ID F1 F2 F3 … Fn
ID Identification number referenced by the NLFREQ field (field 5) of an NLHARM bulk
data entry. (Integer > 0).
Fi Forcing frequency values in cycles per unit time (Real >= 0.0).

Remarks
The frequencies may be specified in any order; they do not have to be in ascending or descending order.
Only one NLFREQ or NLFREQ1 entry with the same ID is allowed.

NLFREQ1
The NLFREQ1 entry is used to define the forcing frequencies for nonlinear harmonic response by using a
start frequency, a frequency interval and a number of intervals.
1 2 3 4 5 6 7 8 9 10
NLFREQ1 ID F1 DF NDF
ID Identification number referenced by the NLFREQ field (field 5) of an NLHARM bulk
data entry. (Integer > 0)
F1 First forcing frequency in the set (real >= 0.0)
DF Frequency increment (real <> 0.0, required). See remark 1.
NDF Number of frequency increments (Integer > 0, Default= 1)

Remarks
1. If DF is positive, it defines a frequency increment. In this case, the first excitation frequency will be
F1 with all subsequent excitations frequencies being of larger value than F1. If DF is negative, it
defines a frequency decrement. In this case, the first excitation frequency will still be F1, but now all
subsequent excitations frequencies will be of smaller value than F1. In both cases, the initial
conditions for a frequency are the response configuration at the previous excitation frequency. The
initial conditions for the first frequency are zero.

Main Index
 188 MSC Nastran Rotordynamics User’s Guide
Analysis Setup

2. Only one NLFREQ or NLFREQ1 entry with the same ID is allowed.

TABLED5
The TABLED5 entry is used generally to define a value as a function of two variables for use in generating
frequency-dependent and time-dependent dynamic loads. In nonlinear harmonic response it is used to define
an NLRGAP whose force-penetration characteristics vary with frequency.
1 2 3 4 5 6 7 8 9 10
TABLED5 TID
X(1) TID(1) X(2) TID(2) X(3) TID(3) X(4) TID(4)
… … ENDT
TID Table identification number. (Integer > 0)
X(i) X value for the function specified by TID(i) (Real, no Default).
TID(i) ID of a TABLED1, TABLED2, TABLED3 or TABLED4 defining the function Y for the
given value of X. (Integer > 0, no Default).

Remarks
1. This table returns a value that is a function of 2 variables . The first variable value x is specified on
this entry as X(i), the function versus y for the specified value for X(i) is specified on the referenced
table TID(i).
2. When used with the NLRGAP entry to define a frequency dependent gap force as a function of
penetration, the frequency values are input as X(i) and the variation of force vs. penetration is input
on a series of TABLED1 entries referenced by the table values TID(i), one for each required change
in properties with respect to frequency.
3. 2D linear interpolation is carried out for values of (x,y) not specified on the tabular entries.

NLRGAP
The NLRGAP entry defines a nonlinear radial (circular) gap for transient response or nonlinear harmonic
response.

NLRSFD
The NLRSFD entry defines a nonlinear radial squeeze film damper for transient response or nonlinear
harmonic response.

Analysis Setup
Setting up a nonlinear harmonic analysis is relatively easy and is a natural progression from a direct frequency
response analysis (SOL 108). It is strongly recommended that you set up the analysis to run linear direct
frequency response and establish a healthy model. Linear CELASi elements along with rigid elements, if
necessary, may be used to connect the degrees of freedom (DOFs) at which the nonlinearities will eventually
be defined. It is also possible at this stage to establish a condensation strategy in order to reduce the number

Main Index
 Chapter 10: Nonlinear Frequency Response 189
Analysis Setup

of degrees of freedom that will be retained for the eventual nonlinear harmonic response analysis. SOL 128
is a completely integrated implementation and it is therefore possible to use any of the condensation methods
usually available for linear and nonlinear analysis. These include:
 Superelements with or without CMS
 A-set definition with or without CMS
 O-set definition with or without CMS
 Automated CMS (MDACMS)
Generally, it is desirable to retain only the DOFs (degree of freedom) connected to nonlinearities for the
resolution of the nonlinear problem, but ultimately the user decides which DOFs to retain. The only
requirement is that the DOFs attached to nonlinearities must be retained; thereafter, any other DOFs may
be added to the retained set. Typically, the number of DOFs connected to nonlinearities is small, even if the
total number of DOFs in the model is large, so using a condensation strategy can yield a significant saving in
elapsed time for the calculation. Condensation may be carried out using static (Guyan) reduction or using
modal synthesis (CMS). To learn more about these methods please refer to the MSC Nastran Superelements User’s
Guide .
Once you have the linear direct frequency response set up, there are only a few simple modifications necessary
to turn this into a nonlinear harmonic response analysis.
1. Change SOL 108 to SOL 128
2. Remove the FREQUENCY= specification in the case control.
In SOL 108, the FREQUENCY= case control command is used to call the set of FREQi entries in
the bulk data which define the set of excitation frequencies for the frequency response problem. In
nonlinear harmonic response, the excitation frequencies are a sequence of frequencies where the initial
conditions of one frequency are the steady state response from the previous excitation frequency.
Consequently, the excitation frequencies are defined on either an NLFREQ or NLFREQ1 bulk data
entry referenced by the NLHARM bulk data entry.
3. Add a case control NLHARM (Case)= entry which references the ID of a bulk data NLHARM entry.
4. Add a NONLINEAR (Case)= case control entry which refers to the set of nonlinear force entries (NOLINi,
NLRGAP, NLRSFD).
5. Remove the modelling technique you used in place of the nonlinear forces.
If CELASi and/or rigid elements were used, these should be removed ready to be replaced by one or
more of the nonlinear force entries NOLINi, NLRGAP or NLRSFD. Note: you may actually wish
to retain some or all of these entries in order to avoid singular structures. This may happen if the
nonlinear force specification results in zero forces between an otherwise unconnected component; in
this case a small spring stiffness between the potentially singular structures can be used to stabilize an
otherwise unstable problem.
6. Add nonlinear force entries at the required degrees of freedom.
There must be at least 1 dof referenced by a nonlinear force entry. The set ID of the collection of
NOLIN1, NOLIN2, NOLIN3, NOLIN4, NLRGAP and NLRSFD entries is called up by the ID of
the case control NONLINEAR= entry.
7. Add a bulk data NLHARM entry.

Main Index
 190 MSC Nastran Rotordynamics User’s Guide
Analysis Setup

This is referenced by the case control NLHARM= entry and defines any sub-harmonic content
(optional), the number of harmonics (required), and the NLFREQ or NLFREQ1 entry which defines
the set of excitation frequencies. See the section above for the format of the NLHARM entry.
8. Add an NLFREQ or NLFREQ1 entry.
If you want to excite the structure at specific frequencies, use the NLFREQ entry. Use the NLFREQ1 entry
to define an increasing or decreasing frequency sequence. See the section above for the format of the
NLFREQ and NLFREQ1 entries.
It is recommended for the first run that the nonlinear force entries define linear characteristics. This will
allow the results to be checked against the linear direct frequency response; the answers should be the same,
even though the problem will be solved using an iterative technique.

Parameters for controlling the analysis

The interface to nonlinear harmonic response is driven by the inputs defined below. The sequence of
excitation frequencies is defined on NLFREQ or NLFREQ1 entries and harmonic response control data
supplied on the NLHARM entry. This allows specification of the number of harmonics and sub-harmonics
to consider for the nonlinear harmonic response analysis as well as referencing the excitation frequency bulk
data entries NLFREQ or NLFREQ1.

PARAM, MXICODE0, 5
In nonlinear harmonic response analysis, if the solution fails to converge more than MXICODE0 times in
succession, a new trial displacement vector is calculated. MXICODE0 allows the number of successive failed
convergences to be modified before a new trial displacement vector is calculated. (Default is 5).

PARAM, NHPLUS, 20
In nonlinear harmonic response analysis, in order to avoid aliasing in the calculation of the Fourier
coefficients, a certain number of extra evaluation points are used. NHPLUS allows the number of extra points
to be defined. (Default is 20).
This technique comes from a paper (Bellinger, 1995) where the number of time steps, S, selected should be
in the range (2K-1) ≤ S ≤ 3(2K-1) and K is the number of frequencies present in the signal.

PARAM, NLHTOL, 1.0E-5

During the iteration procedure of nonlinear harmonic response, the norm of the residual load vector for the
current step is divided by the norm of the residual load vector for the previously converged step; this value is
then compared with NLHTOL. If the value is smaller than NLHTOL, the system is considered to have
converged. (Default is 1.0E-5).

PARAM, NLHTWK, 1.1

In nonlinear harmonic response analysis, if convergence is not obtained, a line search procedure is initiated
to calculate a scaling factor for the displacement vector from which updated nonlinear loads are subsequently
calculated. If the solution fails to converge more than 5 times in succession (modifiable by

Main Index
 Chapter 10: Nonlinear Frequency Response 191
Analysis Setup

PARAM,MXICODE0), a new trial displacement vector is calculated using a push off factor the size of which
is defined by NLHTWK. (Default is 1.1).

Handling Non-Convergence in Nonlinear Harmonic Response

During the ascending or descending sequence of excitation frequencies defined on the NLFREQi entries, if
instability is encountered at a particular frequency, the solution may fail to converge at that frequency. Non-
convergence may be attributed to several causes, and the possibility must always be considered that there
simply is no solution to the system at a particular frequency because the system has become dynamically
unstable.
In most cases of non-convergence not attributed to dynamically unstable conditions, either the number of
harmonics or the sub-harmonic content is insufficient, or the system has reached a bifurcation or turning
point. Try increasing the number of harmonics or sub-harmonics (NLHARM entry) as well as adjusting the
parameters described above. If none of these is successful, the system may have struck a bifurcation point. Try
adjusting the excitation frequencies slightly; for example, if using the following excitation frequencies 5.0,
10.0, 15.0, 20.0…etc., try adjusting these to 5.1, 10.1, 15.1, 20.1. In the current implementation, there is
no automatic treatment for possible bifurcation points and the subject of using a continuation strategy is
under discussion for a future development.
In the case of non-convergence, the response quantities are set to zero, and the calculation continues to the
next excitation frequency in the sequence retaining the initial conditions of the solution from the last
converged frequency. If the solution at the next excitation frequency does not converge, the same procedure
is followed until a converged solution is found.

There may come a time when it is judicious to change the initial conditions particularly when the next
excitation frequency becomes distanced from the last converged frequency after a sequence of failed to
converge excitation frequencies. In this situation, where the initial conditions for an excitation frequency have
become somewhat distanced from the previously converged excitation frequency, it may not mean very much
physically to continue to use the initial conditions from that previously converged excitation frequency.
It would perhaps therefore be favorable to start from zero conditions as always happens for the first frequency
of an analysis starting from scratch. There is some difficulty in deciding how far away from the previously
converged excitation frequency is acceptable to return to zero initial conditions.
No method of resetting initial conditions to zero is programmed as standard, except starting from cold an
analysis with a different frequency specification. For the current implementation, the user must take action
to initiate a strategy.
It is possible that the solution at a particular frequency falls directly on a bifurcation point which creates a
singular problem.

Comments on Nonlinear Forces

As already mentioned, there must be at least one degree of freedom defined on nonlinear force type entries
such as the NLRGAP, NLRSFD and NOLINi entries. If no nonlinear force entries are defined in the input
file, the following message will be issued:

Main Index
 192 MSC Nastran Rotordynamics User’s Guide
Analysis Setup

USER FATAL MESSAGE 9187 (SENLHARM)

THERE ARE NO NONLINEAR DEGREES OF FREEDOM FOR NONLINEAR HARMONIC
RESPONSE ANALYSIS
Nonlinear forces may be defined as functions of displacement or velocity depending on the choice of bulk
data entry. The NLRGAP and NLRSFD entries were designed for defining nonlinear forces in a radial
manner in support of rotating structures, but their use is not limited to these cases. The NLRSFD entry may
be used in conjunction with a user defined service (UDS) to define an arbitrary external function to calculate
forces as a function of several different parameters. The NOLINi entries allow tabular (piecewise) definition
of nonlinear forces (NOLIN1), a scaled product of 2 DOFs (NOLIN2), or a power function in the case of
NOLIN3 and NOLIN4.
For tabular input, the interpolation of nonlinear forces does not undergo any smoothing after the values are
read from the force-displacement definition, so sharp changes in their characteristics can cause convergence
difficulties.

Figure 10-1 Bilinear force interpolation

Figure 10-1 shows a bilinear force-displacement variation for a nonlinear bushing stiffness. In order to avoid a
step change in the definition of the nonlinear force-displacement characteristics, a cubic law may be used to
configure a close match to the bilinear variation using a NOLIN3 and NOLIN4 entry, rather than defining
the bilinear behaviour directly on a NOLIN1 and TABLED1 entry. In some cases, it may not be possible to
do this; in this case, ensure many points are used to define the characteristics on the TABLED1, seeking
gradual changes in characteristics rather than step changes.
Response analysis in the frequency domain involves handling complex quantities to define the real and
imaginary parts of the response. This notion is more easily understood as the magnitude and phase angle of
the response quantities. It must be remembered, however, that nonlinear force definition involves only real
quantities. This is entirely consistent with the way in which the nonlinear harmonic response problem is
solved as the frequency domain (complex) problem is transform to a time domain (real only) problem by
Fourier coefficients in order that an iteration procedure may establish force equilibrium. Once force

Main Index
 Chapter 10: Nonlinear Frequency Response 193
Theoretical Details

equilibrium is established (convergence), the resulting time domain response is then transformed to the
frequency domain using the same Fourier coefficients.

Theoretical Details
Harmonic balance intrinsically uses a linear combination of sinusoidal responses to represent the total
response. If the steady-state response consists of just a few dominant sinusoids, which may easily be the case
if the nonlinearities are mild, then harmonic balance is able to represent the periodic response accurately.
The harmonic balance method converts a system of nonlinear differential equations into a system of
nonlinear algebraic equations, the solution to which is the set of coefficients for the sinusoids making up the
steady-state response. The nonlinear algebraic equations are solved using Newton’s method of iteration.
The general equation of motion reads:

[M ]{u(t )}+ [B]{u(t )}+ [K ]{u(t )} = {P(t )} (1)

where:
[M] System mass
[B] System damping
[K] System stiffness
{u(t)} Displacement at time t
{ u· ( t ) } Velocity at time t
{ u··( t ) } Acceleration at time t
{P(t)} Applied load at time t

To include nonlinear displacement, velocity, and acceleration dependent properties, these can be added as
nonlinear forces {F (u(t ), u (t ), u (t ) )}nl

[M ]{u(t )}+ [B ]{u (t )}+ [K ]{u (t )}+ {F (u(t ), u (t ), u (t ) )}nl = {P (t )} (2)

Equation (2) can be rewritten as,

R{u (t )} = [M ]{u(t )} + [B ]{u (t )} + [K ]{u (t )} + {F (u(t ), u (t ), u (t ) )}nl − {P(t )} = {0} (3)

If the loading and response are periodic, the above equation can be transformed to the frequency domain.

N
(
R{u (ω n )} n =0 = [M ]{u(ω n } + [B ]{u (ω n )} + [K ]{u (ω n )} + Fˆ (ω n ) { }nl
− {P (ω n )} ) N

n =0
= {0}
(4)

Main Index
 194 MSC Nastran Rotordynamics User’s Guide
Theoretical Details

where

ω is 2πf 0 ⁄ ν
f0 is the assumed fundamental frequency of the response
v is the sub-factor for sub-harmonic response
is the Fourier Transform of {F (u(t ), u (t ), u (t ) )}nl
{Fˆ (ω )}
n nl

The presence of nonlinear forces requires an iterative procedure to minimize the function R {u (ω n )} . A
common method to solve this type of problem is Newton’s method, but before applying this procedure,
equation is split into real and imaginary parts and written in matrix format.
Newton’s Method uses the following approximation to attempt to solve nonlinear problems,

∂R{u }
R{u + Δu } = R{u }+ {Δu} = {0}
∂{u }
(5)
or

∂R{u}
{Δu} = − R{u}
∂{u} (6)
The solution uses an iterative approach:
1. The harmonic response is converted to a time-domain response.
2. Nonlinear bearing force derivatives are determined using the responses from step 1.
3. The new bearing force derivatives are used to recalculate the harmonic response.
4. This procedure is repeated until the solution converges.
∂R{u}i
{Δu}i +1 = − R{u}i
∂{u} (7)

where
{u}i+1 = {u}i + {Δu}i+1 (8)

Main Index
 Chapter 10: Nonlinear Frequency Response 195
Examples

Examples

Example 1
This example from (Zu and Ji, 1998) shows a rotor bearing system (see Figure 10-2) in which an out-of-balance
load excites a rigid overhung disk mounted on a flexible shaft turning in bearings exhibiting nonlinear
stiffness.

Figure 10-2 Rotor Bearing System

The round solid shaft of diameter 0.1 metres is 1.0 metre long and runs in two bearings, one located at one
end of the shaft while the other is positioned just inboard of the disk such that the disk is overhung. The
bearings are mounted on an isolation material that exhibits nonlinear stiffness varying with a cubic law. The
overhung massive rigid disk exhibits a small eccentricity in its mass distribution.
Shaft material properties:
Young’s modulus 2.07E+11 Nm-2
Density 7750 kgm-3

Disk properties:
Mass of disk 2000 kg
Inertia (polar) 200 kgm2
Inertia (diametral) 100 kgm2

Bearing stiffness and damping (symmetrical)

Linear stiffness 1.5E+7 Nm-1
Nonlinear stiffness
1E+12d3 Nm-3
(d is the value of displacement)
Damping 1.72E+4 Nsm-1

Main Index
 196 MSC Nastran Rotordynamics User’s Guide
Examples

Firstly, the analysis is run in a linear direct frequency response with linear bearing properties. The analysis is
then repeated in the nonlinear harmonic solution sequence, still with linear bearing properties and the answer
is compared. Finally, the nonlinear bearing stiffness properties are added and the response compared against
theory (Zu and Ji, 1998).

Model Data
The bulk data for the model is shown; this data is common to all runs (common.dat).
PARAM,GRDPNT,0
PARAM,COUPMASS,1
$
$ ROTOR
ROTORG,1,1,2,3,4
RGYRO,66,SYNC,1,FREQ
RSPINR,1,1,4,FREQ,1.
GRID,1
GRID,2,,.5
GRID,3,,1.
GRID,4,,.99
CBEAM,1,1,1,2,,1.
CBEAM,2,1,2,4,,1.
CBEAM,3,1,4,3,,1.
PBEAM* 1 1 7.8539820-3 4.9087390-6
* 4.9087390-6 9.8174780-6
+
.68 .68
MAT1,1,2.07+11,,.27,7.75+3
$
$ DISK MASS & INERTIA
CONM2,6,3,,2000.
,200.,,100.,,,100.
$
$ STATOR
GRID,5
GRID,6,,.99
$
$ ROTOR TO STATOR CONNECTION
RBE2,941,5,123456,1
RBE2,953,6,123456,4
$
$ GROUND
GRID,105
GRID,106,,.99
$
$ BEARING DAMPING
CDAMP2,20442,1.72+4,105,2,5,2
CDAMP2,20443,1.72+4,105,3,5,3
CDAMP2,20552,1.72+4,106,2,6,2
CDAMP2,20553,1.72+4,106,3,6,3
$
$ REMOVE SINGULAR DOFS
SPC1,1,14,2,3,5,6
SPC1,1,123456,105,106

Main Index
 Chapter 10: Nonlinear Frequency Response 197
Examples

$
$ OUT OF BALANCE FOR ROTOR
UNBALNC,77,2000.,3,,1.
,1.4142-3

Linear Frequency Response – SOL 108

In this first analysis, only the linear stiffness of the bearings is considered. This allows easy comparison
between a linear frequency response in SOL 108 and nonlinear harmonic response (with linear nonlinearities)
in SOL 128.
Here is the case control and additional bulk data required for the linear frequency response.
SOL 108
CEND
LINE=9999999
DISP(SORT2,PHASE)=ALL
SPC=1
RGYRO=66
DLOAD=77
FREQ=88
BEGIN BULK
include ‘common.dat’
$
$ BEARING STIFFNESS
CELAS2,10442,1.5+7,105,2,5,2
CELAS2,10443,1.5+7,105,3,5,3
CELAS2,10552,1.5+7,106,2,6,2
CELAS2,10553,1.5+7,106,3,6,3
$
FREQ1,88,9.549296,.0530516,240
ENDDATA

Linear Frequency Response – SOL 128

In this second analysis, the same problem is now run in nonlinear harmonic response which solves the
problem in an iterative manner.
Here is the case control and additional bulk data required for the linear harmonic response in SOL 128.
SOL 128
CEND
LINE=9999999
DISP(SORT2,PHASE)=ALL
SPC=1
RGYRO=66
DLOAD=77
NONLINEAR=1000
NLHARM=2000
BEGIN BULK
include ‘common.dat’
$
$ LINEAR BAERING STIFFNESS (HALVED)
CELAS2,10442,7.5+6,105,2,5,2
CELAS2,10443,7.5+6,105,3,5,3

Main Index
 198 MSC Nastran Rotordynamics User’s Guide
Examples

CELAS2,10552,7.5+6,106,2,6,2
CELAS2,10553,7.5+6,106,3,6,3
$
$ LINEAR BEARING STIFFNESS (HALF PROVIDED BY NOLIN1)
NOLIN1,1000,5,2,-1.,5,2,1001
NOLIN1,1000,5,3,-1.,5,3,1001
NOLIN1,1000,6,2,-1.,6,2,1001
NOLIN1,1000,6,3,-1.,6,3,1001
TABLED1, 1001,
, -1.0, -7.5+6, 1.0, 7.5+6, ENDT
$
NLFREQ1,88,9.549296,.0530516,240
NLHARM,2000,1,1,88
$
ENDDATA

For these two linear analyses, the magnitude response of grid point 3 in Y direction is shown in Figure 10-4.
The linear frequency response curve is hidden by the nonlinear harmonic response curve; this demonstrates
that the linear problem can be solved using the 2 different methods (SOL 108 and 128) and the response is
the same.

Nonlinear Frequency Response – SOL 128

In this third analysis, the cubic nonlinear stiffness of the bearings is added.
Here is the case control and additional bulk data required for the nonlinear harmonic response in SOL 128.
SOL 128
CEND
LINE=9999999
DISP(SORT2,PHASE)=ALL
SPC=1
RGYRO=66
DLOAD=77
NONLINEAR=1000
NLHARM=2000
BEGIN BULK
include ‘common.dat’
$
$ BEARING STIFFNESS
CELAS2,10442,1.5+7,105,2,5,2
CELAS2,10443,1.5+7,105,3,5,3
CELAS2,10552,1.5+7,106,2,6,2
CELAS2,10553,1.5+7,106,3,6,3
$
$ CUBIC NONLINEAR STIFFNESS (TENSION)
NOLIN3,1000,5,2,-1+12,5,2,3.
NOLIN3,1000,5,3,-1+12,5,3,3.
NOLIN3,1000,6,2,-1+12,6,2,3.
NOLIN3,1000,6,3,-1+12,6,3,3.
$ CUBIC NONLINEAR STIFFNESS (COMPRESSION)
NOLIN4,1000,5,2,-1+12,5,2,3.
NOLIN4,1000,5,3,-1+12,5,3,3.

Main Index
 Chapter 10: Nonlinear Frequency Response 199
Examples

NOLIN4,1000,6,2,-1+12,6,2,3.
NOLIN4,1000,6,3,-1+12,6,3,3.
$
$ Spin up
NLFREQ1,88,9.549296,.1591549,240
$ Spin down
$NLFREQ1,88,47.74648,-.1591549,240
NLHARM,2000,1,1,88
$
ENDDATA

The analysis is run in two parts, the first starting from a cyclical frequency of 9.55 Hz (60 radians/second)
with an increasing frequency up to a frequency of 47.7 Hz (300 radians/second); this simulates a spin-up
event where each frequency is considered in its steady state condition. The second part of the analysis
simulates a spin-down event starting from a cyclical frequency of 47.7 Hz with a decreasing frequency.
The magnitude of the Y direction response of GRID point 3 is plotted against rotational frequency with the
resulting two curves:

Figure 10-3 The Y direction response of GRID point 3

Clearly there is a zone of bifurcation just before 23 Hz; that is to say, after 23 Hz, two possible states exist. In
the spin-down case, the solution jumps from one solution to the other and then retraces the spin-up response
curve; an unstable condition exists between these two.
The results are in good agreement with those reported in reference (Zu and Ji, 1998).

Main Index
 200 MSC Nastran Rotordynamics User’s Guide
Examples

Example 2
Use of NLRGAP element
The NLRGAP element can be used to define nonlinear force-displacement relation as a function of frequency
for nonlinear frequency response analysis performed using SOL 128. Brief description of this feature is given
in QRG description of NLRGAP element.
For nonlinear harmonic response either a TABLED1 entry or a TABLED5 entry may be referenced leading
to three possible scenarios:
a. The number entered in the TABK field is a positive integer referencing a TABLED1 entry. In this
case, the pairs of values defined on the TABLED1 entry are frequency vs. gap stiffness.
b. The number entered in the TABK field is a negative integer, the absolute value of which references
a TABLED1 entry. In this case, the pairs of values defined on the TABLED entry are gap
penetration vs. gap force.
c. The number entered in the TABK field is a positive integer referencing a TABLED5 entry. In this
case, the pairs of values defined on the TABLED5 entry are frequency vs. a TABLED1 ID that
defines pairs of gap penetration vs. gap force.
Example: Consider the NM rotor used earlier in the analysis.

Figure 10-4 NM rotor used in the analysis

The rotor used in this example problem is shown in Figure 10-4. The unbalance load is defined on the Grid
41 while the rotor is supported using linear bearings at Grid 91 and Grid 111. For this analysis, additional
nonlinear bearing elements are defined at Grid 91. Nonlinear element used in this analysis is NLRGAP and
it defines nonlinear force-deflection relationship which varies with frequency, see Figure 10-5. Changes made
in the input file to introduce the nonlinear bearing element is described below:

Main Index
 Chapter 10: Nonlinear Frequency Response 201
Examples

Figure 10-5 Nonlinear Force deflection curve as a function of frequency

For this particular case, introduction of the nonlinear NLRGAP elements results in an increase in the stiffness
of support structure. As a result, the peak frequency is shifted to the right, as seen in Figure 10-6.

Main Index
 202 MSC Nastran Rotordynamics User’s Guide
Examples

Figure 10-6 Effect of nonlinearity

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 10-1
Input File Name Description
rug_10_1 Linear response
rug_10_2 Linear response in SOL 128
rug_10_3 Nonliner response in Spin up.
rug_10_4 Nonlinear response in spin down.
rug_10_5 NM model with nonlinear NLRGAP element.

Main Index
 Chapter 11: Superelement Analysis with Rotors
MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Superelement Analysis with

11 Rotors


Introduction 204
 Support for Copy and Move/Mirror Capability 205

Sample Problem 207

Results from External SE Analysis 215

Main Index
 204 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
The principle used in superelement (SE) analysis is often referred to as substructuring. That is, the model is
divided into a series of components, each of which is processed independently resulting in a set of matrices
that are reduced to a boundary and describe the behavior of the component as seen by the rest of the structure.
Often these components are comprised of logical groupings of elements (an engine, a wing, a fender, the
exhaust system, etc.), hence the term superelement.
There are three methods available for defining superelements:
 List Superelement: Commonly referred to as “SESET” superelements
 PART Superelement: Commonly referred to as “BEGIN SUPER” superelements
 External Superelement: Commonly referred to as “EXTSEOUT” or “EXTOUT” or “DMIG”
superelements
Of the three available methods, rotors can be defined in PART or External SEs

PART Superelements
PART superelements are defined by defining each superelement in its own Partitioned Bulk Data section.
These separate sections of the bulk data are self-contained in that each section contains all geometry,
elements, properties, constraints, parameters, and loading data for that component of the model. When
PARTs are used the program works in a manner similar to an assembly process. That is, a series of separate
components are assembled into the final finite element model, i.e. the residual structure.

External Superelements
The external superelement (SE) capability involving the use of the EXTSEOUT Case Control command was
added to MSC Nastran in 2004. Since then, this capability has become the most popular and most widely
employed SE usage. External Superelements are similar to PART superelements in many respects, except
rather than solving the model in a single run, the superelement can be processed and output for use at a later
time. There are many advantages of external superelements: 1) the reduced matrices are compact and can be
added to another structure while maintaining full fidelity of the component behavior on the system, 2) they
can be easily re-used as many times as necessary at a very low runtime cost, 3) they can protect design
information (proprietary geometry) and material information (composite layup), 4) key results can be
monitored without the need for full data recovery, 5) files can be easily shared and maintained across different
organizations or design groups.
Support for rotors in external SE was implemented in V2014, while support for rotors in PART SE is
introduced in V2018. With these enhancements in MSC Nastran, any number of ROTORG, ROTOR (in
fixed reference) and/or ROTORAX Bulk Data entries that define rotors may be specified in external or part
SEs and the residual, but Bulk Data entries that pertain to usage of rotors (like RGYRO, RSPINR, RSPINT
and UNBALNC entries) are allowed and processed only in the residual.
This design has the advantage that, once the rotors are defined in the SEs, different variations of their usage
may be accomplished in the assembly job by having different variations of the RGYRO, RSPINR and/or
RSPINT rotor usage entries. Thus, this design gives the user the freedom to form different Rotordynamic

Main Index
 Chapter 11: Superelement Analysis with Rotors 205
Support for Copy and Move/Mirror Capability

configurations in assembly jobs for performing specific types of analysis like complex eigenvalue analysis or
frequency or transient response analysis.
With the above design, gyroscopic matrices and other rotor related matrices are computed for rotors in the
SE in which they are defined (whether it be an external SE, part SE or the residual). Speed factors are applied
to rotor related matrices of all rotors only in the residual. All of these matrices are then combined
appropriately to perform the analysis specified by the user.
The above scenario also introduces another interesting aspect into the design. With the enhancements in
MSC Nastran 2014, it is quite possible that there may very well be rotors with duplicate IDs across external
and part SEs and the residual. In order to allow for this scenario, rotors whose usage is specified in the
assembly job need to be identified not only by their rotor IDs, but also by the IDs of the SEs in which they
are defined. To facilitate this, a field called ROTRSEID has been added to the existing RGYRO, RSPINR
and RSPINT Bulk Data entries to specify the ID of the SE in which a particular rotor is defined.
Yet another point to consider is that the mass summary information for rotors is generated and output only
in the jobs that define the rotors. Thus, for instance, if there are rotors in both SEs and the residual, the
assembly job generates and outputs rotor mass summary information only for the rotors defined in the
residual. The mass summary information for the rotors defined in upstream external SEs is available only in
the corresponding external SE creation jobs. The program points this out via an appropriate user information
message in the assembly job if any rotors defined in external SEs are referenced in the assembly job.

Support for Copy and Move/Mirror Capability

There are many practical cases in which we encounter geometrically identical components. Blades of turbines
and engines and many components in airplanes and automobiles are examples of such cases.
When external SEs are employed to represent such components, the copy and move/mirror feature available
since MSC Nastran 2014 allows the user to employ additional copies of an already reduced external SE (the
“primary” SE) in an assembled configuration without having to perform reduction operations on each of the
additional geometrically identical components (the “secondary” SEs). Instead, the boundary matrices for all
of the “secondary” external SEs are obtained automatically via appropriate internal transformations of the
corresponding boundary matrices of the “primary” external SEs. The translation and rotation of element
coordinate system during the copy and mirror procedure is shown in Figure 11-1.
The following important points should be noted with regard to the usage of the copy and move/mirror
feature:
 The primary SE does not include SEs upstream of the primary SE.
 If the primary SE is a part SE, then the secondary SE is a "G-set" copy of the primary SE. In this
case, the boundary, loads, constraints and reduction procedure of a secondary SE can be different
from those of its primary SE.
 If the primary SE is an external SE resulting from the use of the EXTSEOUT Case Control
command in an earlier job, then the secondary SE is an "A-set" copy of the primary SE. In this case,
the boundary, loads, constraints and reduction procedure of the secondary SE are set and are the
same as those of its primary SE.

Main Index
 206 MSC Nastran Rotordynamics User’s Guide
Support for Copy and Move/Mirror Capability

 A secondary SE requires the specification of either an SELOC entry or an SEMPLN entry. If an

SELOC entry is specified, then an identical copy of its primary SE will be positioned at the location
implied by the SELOC entry. If an SEMPLN entry is specified, then a mirror image of the primary
SE will be positioned.
 It should be noted that even a primary external SE can be re-positioned in an assembly run by use of
an SELOC or SEMPLN entry.
The above capability greatly enhances user convenience and productivity by eliminating the need to
individually model geometrically identical external SEs and contributes to enhanced efficiency for
performing the analysis of such assembled configurations.

Figure 11-1 Copy/Move feature in MSC Nastran for superelements

Main differences in the implementation for supporting rotors in the external SE and the part SE are as
follows:

Table 11-1 Differences in External SE and PART SE implementation

Part SE External SE
Multilevel SE Rotors can be defined in any of the Rotor can only be defined one level
SEs in a multilevel SE model below the residual.
Grids on RSPINR/T Grids defined on RSPINR/RSPINT Grids defined on RSPINR/RSPINT
need not be part of ASET for that SE MUST be part of ASET for that SE
QSET for secondary SEs For a secondary SE generated using For a secondary SE generated using
mirror/move option, QSET DOFs mirror/move option, all the DOFs
must be redefined to include their are defined automatically.
effect in the analysis.

Main Index
 Chapter 11: Superelement Analysis with Rotors 207
Sample Problem

Sample Problem
The sample problem consists of a model plane with two identical engines is analyzed where each engine
includes two axisymmetric rotors. For this case, rotors are defined in SEs one level below the residual. Thus,
this problem can be solved using any of the external SE or PART SE approaches. Here only the external SE
approach is demonstrated, however, similar steps can be used for PART SE approach too.
In order to compare the results, three different cases are considered as indicated below.
a. Single-shot run: Here, the complete plane with both the engines is analyzed at once, without the
usage of external superelements (see Figure 11-2)
b. External SE assembly run using two external SEs: In this case, the left and right engines are part
of two separate external superelements with IDs of 200 and 400, respectively. The model used for
the creation run of the right engine is shown in Figure 11-3. The model includes both rotating and
non-rotating components. Besides the physical DOFs of the boundary points shown in
Figure 11-3, modal DOFs for the complete model are also used in the analysis performed in the
assembly run.
The model for the assembly run with the left and right engines included as external superelements
is shown in Figure 11-4.
c. External SE assembly run using one external SE and the MIRROR feature: In this case, only the
external SE model for the right engine is used. The left engine is attached to the residual structure
by mirroring the right engine about the mirror plane as shown in Figure 11-5.

Main Index
 208 MSC Nastran Rotordynamics User’s Guide
Sample Problem

Figure 11-2 Model for single-shot run

Figure 11-3 Engine model used for external SE creation run

Main Index
 Chapter 11: Superelement Analysis with Rotors 209
Sample Problem

Figure 11-4 Model with engines as two external SEs

Figure 11-5 Model with left engine obtained using mirror feature

Input File for Creation Run

Both SOL 101 and SOL 103 can be used for the creation run. Additional features required in different
sections of the input file for creating external SE is described below:

File Management
$ ASSIGN STATEMENTS FOR STORING EXTERNAL MATRICES IN THE OP2 FILE
ASSIGN OUTPUT2='freedom_103a_op2' UNIT=35 DELETE

Case Control
$ INCLUDE THE EXTSEOUT COMMAND
extseout(asmbulk=auto, extbulk, extid=200, dmigop2=35)

Main Index
 210 MSC Nastran Rotordynamics User’s Guide
Sample Problem

Bulk data Entries

Figure 11-6 ASET and QSET DOFs for the creation run

Main Index
 Chapter 11: Superelement Analysis with Rotors 211
Sample Problem

Successful execution of the input file for creation run generates following files at output
a. *.asm File
This file identifies SE in the model and ID corresponding OP2 file. This file can be included anywhere in
the assembly run. Snapshot of the file is shown in Figure 11-7.

Figure 11-7 Snapshot of the *.asm file

Main Index
 212 MSC Nastran Rotordynamics User’s Guide
Sample Problem

b. *.pch File
The PCH files includes all the bulk data entries pertaining to the external SE. It identifies external boundary
points, and lists out grid location of all the ASET DOFs to be used in the assembly run. Information
regarding modal DOFs is also included in this analysis. Snapshot of the PCH file is shown in Figure 11-8

Figure 11-8 Snapshot of the PCH file

c. OP2 file
This file contains all the matrices generated during the creation run.

Assembly Run
The analysis for the assembled structure with all the external SEs created earlier can be performed in any of
the solution sequences where rotors are supported. In order to include an external SE with rotor in the
residual run, following changes need to be made in the assembly run:

File Management section

$ ASSIGN STATEMENTS FOR READING EXTERNAL MATRICES FROM THE OP2 FILE
ASSIGN INPUTT2='freedom_103a_op2' UNIT=35 $

Main Index
 Chapter 11: Superelement Analysis with Rotors 213
Sample Problem

Bulk data Section

Copying an External SE
Copying an external SE created apriori and then relocating it to a different location in the model requires
usage of the SEBULK and SELOC bulk data entries.
SELOC Partitioned Superelement Location

Defines a partitioned superelement relocation by listing three noncolinear points in the superelement and
three corresponding points not belonging to the superelement.

Format:
1 2 3 4 5 6 7 8 9 10
SELOC SEID PA1 PA2 PA3 PB1 PB2 PB3

Figure 11-9 Snapshot of SELOC entry

In the example shown below, a new external SE with SE ID 4 is created by repeating SE with ID 3. And then
it is repositioned in the model by defining three points (19, 20, and 36) that belong to the SE and three points
(1027, 1028, and 1042) that are part of residual model.
SEBULK,4,REPEAT,3
SELOC,4,19,20,36,1027,1028,1042

Generating mirror image of an external SE

A mirror image of an existing external SE can be created by using combination of SEBULK and SEMPLN
entries. Brief description of the SEMPLN entry is given in Figure 11-10. It requires the user to list three grids
or points that define the plane of symmetry. The user needs to ensure that there exists corresponding grid

Main Index
 214 MSC Nastran Rotordynamics User’s Guide
Sample Problem

points in the residual model to make the connection between the new external SE and residual. This
connection can be verified by looking at the output in.F06 file, as shown in Figure 11-11.
SEMPLN Superelement Mirror Plane

Defines a mirror plane for mirroring a partitioned superelement.

Format:
1 2 3 4 5 6 7 8 9 10
SEMPLN SEID “PLANE” P1 P2 P3

Figure 11-10 Snapshot of the SEMPLN entry

In the example shown below, a mirror image of an existing SE with ID 400 is created using YZ as the mirror
plane.

sebulk,200,repeat,400
sempln,200,plane,800001,800002,800003
POINT 800001 0.00.00.0
POINT 800002 1.00.00.0
POINT 800003 0.00.01.0

Information in the Output File

Besides the actual output requested in case control, the external SE assembly run provides information about
the boundary connection and rotors defined in the external SEs, as shown in Figure 11-11 and Figure 11-12,
respectively. Figure 11-11 shows the connection made between the grid points defined in the assembly
(residual) run and grid points defined as part of ASET in the external superelement creation run. It also lists
all the modal DOFs defined in the creation run. User need to ensure that correct connection has been made
between the grid points defined in residual and in EXT SE to obtain correct results. In case rotors are present
in the EXT SE creation run, the .F06 file for the assembly run lists all the rotors defined in the external
superelements as shown in Figure 11-12. Other details related to rotor mass and spin direction is available in
the .F06 file of the creation run.

Main Index
 Chapter 11: Superelement Analysis with Rotors 215
Results from External SE Analysis

Figure 11-11 Boundary connection information in .F06 file

Figure 11-12 Rotor information in .F06 file

Results from External SE Analysis

In order to validate the external Superelement analysis, three different types of models described previously
are analyzed using:
i) SOL 107 to obtain complex eigenvalue frequencies;
ii) SOL 108 to obtain frequency response for external applied force, and
iii) SOL 109 to obtain direct transient response.

Main Index
 216 MSC Nastran Rotordynamics User’s Guide
Results from External SE Analysis

SOL 107 Results (Complex Eigenvalue Analysis)

The Campbell diagram obtained for the three cases, shown in Figure 11-13, Figure 11-14 and
Figure 11-15, are similar.

Figure 11-13 Campbell diagram for single-shot run (without tracking)

Main Index
 Chapter 11: Superelement Analysis with Rotors 217
Results from External SE Analysis

Figure 11-14 Campbell diagram for the external SE assembly run with two external SEs (without tracking)

Figure 11-15 Campbell Diagram for external SE assembly run using mirror feature (without tracking)

Further comparison is performed for frequencies 1, 5, 9 and 13, as shown in Figure 11-16. The results obtained
in this section demonstrate that the complex eigenfrequencies obtained for each of the three cases are similar,
thus validating the implementation for the external SE analysis.

Main Index
 218 MSC Nastran Rotordynamics User’s Guide
Results from External SE Analysis

Figure 11-16 Comparison for different frequencies (with mode tracking)

SOL 108 Results (Direct Frequency Response)

For this analysis, a force of fixed amplitude at varying frequency (same as the rotational speed of the rotor) is
applied to the rotors in the engine and the response at the tip of the right wing and tip of the tail is observed.
The results shown in Figure 11-17 and Figure 11-18 show good agreement between the results obtained for the
three cases.

Main Index
 Chapter 11: Superelement Analysis with Rotors 219
Results from External SE Analysis

Figure 11-17 Displacement at the tip of the right wing

Figure 11-18 Displacement at the tip of the tail

Main Index
 220 MSC Nastran Rotordynamics User’s Guide
Results from External SE Analysis

SOL 109 Results (Direct Transient Response)

As in the previous case, time-dependent forces of fixed amplitude and frequency is applied to the rotors in
the engine whose rotational speed is increasing linearly with time and the response at the tip of the right wing
and tip of the tail is observed. The results obtained for direct transient response match very well for the three
cases considered here (see Figure 11-19 and Figure 11-20).

Figure 11-19 Displacement at the tip of Right wing

Figure 11-20 Displacement at the tip of the tail

Main Index
 Chapter 11: Superelement Analysis with Rotors 221
Multilevel SE Setup

Multilevel SE Setup
This feature is supported for rotors defined in PART SE only. The user interface for this feature is similar to
the existing approach used for standard Part SEs and for defining rotors in External SE. Here, the rotor is
defined in the BEGIN SUPER section of the model and corresponding RSPINR/RSPINT entries and
RGYRO entry are defined in the MAIN BULK DATA section of the deck. Unlike external superelements,
the grids specified on the RSPINR/RSPINT need not be part of the ASET for part SE rotor is defined in.
However, if unbalance loads need to be applied on any of the rotors in upstream superelements, then the grid
where unbalance loads are defined, MUST be part of the ASET and should have a corresponding connection
point in the residual.
A hypothetical test case is shown in Figure 11-21 where multiple rotors are defined in multilevel PART SE
setup.

Rotor and QSET definition in SE 442

BEGIN SUPER=442
ROTORG 12 25 26 27 28
ROTORG 15 51 52 53 54
QSET1 0 6001 THRU 6010
SPOINT 6001 THRU 6010
QSET definition in SE 441

QSET1 0 5001 THRU 5015

SPOINT 5001 THRU 5014
SPOINT 16001 THRU 16010

Rotor and QSET definition in SE 88

BEGIN SUPER=88
ROTORG 4 21 22 23 200
ROTORG 14 41 42 43 423 44
QSET1 0 2001 THRU 2012
SPOINT 2001 THRU 2012
Rotor and QSET definition in SE 44

Main Index
 222 MSC Nastran Rotordynamics User’s Guide
Multilevel SE Setup

BEGIN SUPER=44
ROTORG 5 17 THRU 19
ASET 140 123456 18 123456
QSET1 0 1001 THRU 1016
SPOINT 1001 THRU 1016
SPOINT 12001 THRU 12012
SPOINT 15001 THRU 15014
Rotor definition and SE connection in Residual

DTI SETREE 1 442 441 441 44 88 44

44 0
ROTORG 6 7 THRU 12
SECONCT 44 0
140 14 1001 THRU 1016 11001 THRU 11016
18 18
SPOINT 11001 THRU 11016

SECONCT 88 44
200 20 2001 THRU 2012 12001 THRU 12012

SECONCT 441 44
5001 THRU 5012 15001 THRU 15012 220 20

SECONCT 442 441

6001 THRU 6010 16001 THRU 16010 25 250

RSPINR/RSPINT definition in residual

RSPINR 4 21 22 RPM 1. 88
RSPINR 14 41 42 RPM 1. 88
RSPINR 5 18 19 RPM 1. 44
RSPINR 12 25 26 RPM 1. 442
RSPINR 15 51 52 RPM 1. 442
RSPINR 6 7 12 RPM 1.
SEs are analyzed based on the process order determined using description provided in SETREE.

0 SUPERELEMENT DEFINITION TABLE (SORTED BY SEID)

0 PRIMARY PROCESS DOWNSTREAM
SUPERELEMENT SUPERELEMENT ORDER SUPERELEMENT TYPE LABEL
------------ ------------ ------- ------------ ----------------------------- -------------------------
--------------------
0 0 5 0 RESIDUAL STRUCTURE
44 0 4 0 PRIMARY (BEGIN SUPER)
88 0 2 44 PRIMARY (BEGIN SUPER)
441 0 3 44 PRIMARY (BEGIN SUPER)
442 0 1 441 PRIMARY (BEGIN SUPER)
Mass summary for each rotor is produced as the SE it is defined in is analyzed. Final rotor spin direction for
each rotor is determined based on the grid order prescribed in RSPINR/RSPINT entries.

Main Index
 Chapter 11: Superelement Analysis with Rotors 223
Multilevel SE Setup

* USER INFORMATION MESSAGE 2064 (ROTOR2)

FOR ROTOR WITH ID 4 AND SEID 88, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
* USER INFORMATION MESSAGE 2064 (ROTOR2)
FOR ROTOR WITH ID 5 AND SEID 44, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
* USER INFORMATION MESSAGE 2064 (ROTOR2)
FOR ROTOR WITH ID 6 AND SEID 0, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
* USER INFORMATION MESSAGE 2064 (ROTOR2)
FOR ROTOR WITH ID 12 AND SEID 442, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
* USER INFORMATION MESSAGE 2064 (ROTOR2)
FOR ROTOR WITH ID 14 AND SEID 88, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
* USER INFORMATION MESSAGE 2064 (ROTOR2)
FOR ROTOR WITH ID 15 AND SEID 442, SPIN DIRECTION IS SAME AS THE
ASSUMED DIRECTION
Results obtained using PART SE approach are compared with those obtained without SE.
Results at rotor speed of 10000 RPM

Table 11-2 Complex eigenvalues for multilevel Part SE case

Mode Number NO PS (Hz) Mixed (Hz) RotorG (Hz)
1 0.461 0.461 0.461
2 0.461 0.461 0.461
3 0.485 0.485 0.485
4 0.485 0.485 0.485
5 0.869 0.869 0.869
6 0.869 0.869 0.869
7 1.302 1.302 1.302
8 1.302 1.302 1.302
9 3.761 3.763 3.763
10 3.761 3.763 3.763
11 4.029 4.030 4.030
12 4.029 4.030 4.030
13 6.128 6.132 6.132
14 6.128 6.132 6.132
15 6.538 6.543 6.543
16 6.538 6.543 6.543

Main Index
 224 MSC Nastran Rotordynamics User’s Guide
Multilevel SE Setup

Table 11-2 Complex eigenvalues for multilevel Part SE case

Mode Number NO PS (Hz) Mixed (Hz) RotorG (Hz)
17 9.773 9.782 9.782
18 9.773 9.782 9.782
19 10.847 10.858 10.858
20 10.847 10.858 10.858

Examples used in this chapter are the same as those used in the v2014 release guide. These examples are
available in the following folder: ../tpl/extsecopy/ and ../tpl/rotv18/

Table 11-3
Input File Name Description
freedom_103a.dat Creation run for the left engine
freedom_103b.dat Creation run for the right engine
freedom_107a.dat Assembly run for model with two ext SEs in SOL 107
freedom_107b.dat Assembly run for model with mirrored ext SE in SOL 107
freedom_107s.dat Single-shot run for freedom plane in SOL 107
freedom_108a.dat Assembly run for model with two ext SEs in SOL 108
freedom_108b.dat Assembly run for model with mirrored ext SE in SOL 108
freedom_108s.dat Single-shot run for freedom plane in SOL 108
freedom_109a.dat Assembly run for model with two ext SEs in SOL 109
freedom_109b.dat Assembly run for model with mirrored ext SE in SOL 109
freedom_109s.dat Single-shot run for freedom plane in SOL 109
../tpl/rotv18/r1d10 Multiple Rotors in residual (No PART SE case)
7nops_18.dat
../tpl/rotv18/rbd10 Rotors defined using ROTOR and ROTORG in multilevel PART SEs
7mlps_19.dat
../tpl/rotv18/r1d10 Rotors defined using ROTORG in multilevel part SE
7mlps_20.dat

Main Index
 Chapter 12: Multiple Rotors In Fixed System

MSC Nastran Implicit Nonlinear (SOL 600) User’s Guide

Multiple Rotors In Fixed

12 System

 Introduction 226

Main Index
 226 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
MSC Nastran supports multiple rotors in an analysis model. They can be all of the same type or different
(ROTORG, ROTORAX, or ROTOR). (In case, the ROTOR entry is used for analyzing a rotating structure
in rotating reference frame, then ROTORAX and ROTORG cannot be used since they are defined in fixed
reference frame.)
For analysis in the fixed reference frame, rotor speed of each rotor can be specified independently. User has
the option to specify rotor speed as a real multiple of the reference rotor`s speed or using a table lookup. For
complex eigenvalue analysis and for frequency response analysis, both asynchronous and synchronous
analysis are supported. For synchronous analysis in SOL 107 and synchronous analysis in SOL 108 with the
GYROAVG option, the rotor speed of each rotor is approximated as linear function of reference rotor`s
speed. For all other cases, actual rotor speed of each rotor is used in the analysis. (See further details in QRG
under the RSPINR entry, or equations in Chapter 7 and Chapter 8)

Example
For the purpose of demonstration, consider a second cantilever rotor defined next to a NM rotor as shown
in Figure 12-1. Here, both the rotors are defined using ROTORG and they are completely independent of each
other.
NM Rotor

Cantilever Rotor

Figure 12-1 Rotor models used for analysis

The second rotor defined above is analyzed first separately using SOL 103 and SOL 107 to determine real
eigenvalues and critical eigenvalues, respectively. The results obtained are shown in Table 12-1.

Main Index
 Chapter 12: Multiple Rotors In Fixed System 227
Introduction

Table 12-1 Real and critical eigenvalue for cantilever rotor (Hz)

Table 12-2
Mode Real Eigenvalues Critical Frequencies
1 189.91 188.47
2 189.91 191.38
3 1152.22 1088.12
4 1152.22 1222.50
5 1236.60 1236.60

Steps to introduce multiple rotors in the analysis

For each rotor defined in the analysis, corresponding rotor definition card and RSPINR entry need to be
defined. Also, only one RGYRO entry is required in the analysis which identifies reference rotor and rotor
speed for asynchronous analysis.
Case 1: Rotors spinning at same speed
RGYRO 1 SYNC 1 RPM 1000.0
$---------------- ROTOR 1 ---------------------------------------
$ Rotor Dynamics data
ROTORG 1 1 THRU 15
ROTORG 1 41 91 111
RSPINR 1 9 10 RPM 1.
$---------------- ROTOR 2 ------------------------------------------
ROTORG 2 21 THRU 26
RSPINR 2 21 22RPM 1.
In this model, both Rotor 1 and Rotor 2 are defined with Rotor 1 as the reference rotor. When complex
eigenvalue analysis is performed for this model with the SYNC option, following critical speeds are obtained:

Main Index
 228 MSC Nastran Rotordynamics User’s Guide
Introduction

Table 12-3 Critical Frequencies for model with multiple rotors

Table 12-4
Mode Frequencies (Hz) Source
1 189.91 Rotor 2
2 189.91 Rotor 2
3 270.30 Rotor 1
4 270.30 Rotor 1
5 795.15 Rotor 1
6 795.15 Rotor 1
7 1025.38 Rotor 1
8 1152.22 Rotor 2
9 1152.22 Rotor 2
10 1236.60 Rotor 2

Here, the MSC Nastran results do not identify corresponding dominant rotor for the critical frequencies
obtained. However, MSC Nastran provides rotor energies for each mode as shown in Figure 12-2. Based on
this information, the dominant rotor for each critical frequency was identified as shown in Table 12-3 .

Main Index
 Chapter 12: Multiple Rotors In Fixed System 229
Introduction

Figure 12-2 Rotor energy distribution

In case the rotors are spinning at different rotor speeds, there are two options currently available in MSC
Nastran to provide different rotor speeds as a function of reference rotor spin speed. These options are shown
below:
Case 2: Rotors spinning at different speed
 Option 1
RSPINR 1 9 10 RPM 1.
RSPINR 2 21 22 RPM 2.
 Option 2

RSPINR 1 9 10 RPM 11
DDVAL 11 100. 200.
RSPINR 2 21 22 RPM 22
DDVAL 22 200. 400.

Main Index
 230 MSC Nastran Rotordynamics User’s Guide
Introduction

In this first option, the speed of second rotor is direct multiple of reference rotor speed. However, in case
option 2, any generic piecewise linear variation can be specified. Critical frequencies obtained for this case
are shown in Table 12-5 and compared with those obtained earlier in Table 12-3. Note that the critical
frequencies of the second rotor have shifted slightly in this case. This is due to the fact the synchronous
analysis is now looking for the frequencies where the rotor speed is half of the structural frequency. Critical
speeds for the reference rotor is the same both cases.

Table 12-5 Results from SOL 107 analysis (SYNC option)

Case 1 Case 2
Mode Frequencies (Hz) Source Frequencies (Hz)
1 188.47 Rotor 2 187.05
2 191.38 Rotor 2 192.88
3 257.31 Rotor 1 257.31
4 284.75 Rotor 1 284.75
5 767.55 Rotor 1 767.55
6 820.14 Rotor 1 820.14
7 1025.38 Rotor 1 1025.38
8 1088.12 Rotor 2 1030.65
9 1222.50 Rotor 2 1088.70
10 1236.60 Rotor 2 1236.60

For the analysis with multiple rotors spinning at different speeds, correct critical frequencies for Rotor 2 can
be obtained by performing the synchronous complex eigenvalue analysis again with Rotor 2 as the reference
rotor.
RGYRO 1 SYNC 2 RPM 1000.0
RSPINR 1 9 10 RPM 1.
RSPINR 2 21 22 RPM 2.
Results obtained for this case are shown in Table 12-6. It should be noted that in this case, the critical
frequencies obtained for Rotor 1 will be incorrect.

Main Index
 Chapter 12: Multiple Rotors In Fixed System 231
Introduction

Table 12-6 Reference Rotor: Rotor2

Table 12-7
Reference Rotor: Rotor2
Mode Frequencies (Hz) Source
1 188.47 Rotor 2
2 191.38 Rotor 2
3 263.63 Rotor 1
4 277.34 Rotor 1
5 781.69 Rotor 1
6 807.96 Rotor 1
7 1025.38 Rotor 1
8 1088.12 Rotor 2
9 1177.44 Rotor 1
10 1222.50 Rotor 2

Complex Frequency Response analysis

MSC Nastran allows the user to specify multiple unbalance loads. These loads can be specified on the same
rotor or on different rotors. Unbalance loads are currently determined based on the reference rotor speed.
Multiple unbalance loads can be specified in the bulk data section in following way:
$ Multiple Unbalance Loads
$ For ROTOR 1
RSPINR 1 9 10 RPM 1.
UNBALNC 10 122.68 41 1. 0. 0.
1.-4 0.0 0.0 0.0 1000.
$ For ROTOR 2
RSPINR 2 21 22 RPM 1.
UNBALNC 10 122.68 26 1. 0. 0.
1.-4 0.0 0.0 0.0 1000.
Note that both the unbalance entries have same ID which is referenced by the DLOAD entry in CASE
Control. Also, the user does not need to specify the rotor ID for unbalance loads explicitly. MSC Nastran
automatically identifies rotor for applying unbalance load based on the grid point specified. Unbalance loads
generated due to above bulk data entries and corresponding displacement is shown in Figure 12-3 and
Figure 12-4, respectively. The results of displacement in Figure 12-4 show peaks at the forward whirl critical
speed for each of the rotor, as obtained earlier in Table 12-3.

Main Index
 232 MSC Nastran Rotordynamics User’s Guide
Introduction

Figure 12-3 Unbalance loads on the two rotors

Figure 12-4 Displacement due to unbalance loads (undamped case)

Transient Analysis
MSC Nastran supports transient analysis with multiple rotors operating at different rotor speeds. In this
analysis, no approximation is made for determining the rotor speeds. Multiple unbalance loads or any kind

Main Index
 Chapter 12: Multiple Rotors In Fixed System 233
Introduction

of external loads can be applied to the rotating or non-rotating components of the model. Changes to be
made in the input file to model multiple rotors are described below.

Case Control Section

In case control, the RYGRO entry is specified. It also includes set of EPOINTS which store force output from
the unbalance loads and variation of rotor speed with time. Output from EPOINTS can be extracted using
SDISP.
RGYRO = 10
TSTEP = 1
SET 99 = 41,26
SET 98 = 92,93,94,95,96,97,98,99
DISP(SORT2,PHASE)=99
OLOAD(SORT2,PHASE)=99
NLLOAD(SORT2,PHASE)=99
SDISP(SORT2) = 98

Bulk Data Section

Rotor speed for each of the rotors can be specified independently through the RSPINT entry. It also includes
the option to define an EPOINT which can be used to save variation of rotor speed with time at each time
step. In this case, EPOINT 95 and 96 are used to save rotor speed information for rotor 1 and rotor 2
respectively.
$ Rotors Relative Spin Rates
RSPINT 1 9 10 FREQ 1 95
TABLED1 1
0.0 0.0 0.005 0.0 .5 400. 1000. 400.
ENDT
RSPINT 2 21 22 FREQ 2 96
TABLED1 2
0.0 0.0 0.005 0.0 .5 800. 1000. 800.
ENDT
Unbalance loads are specified in the similar manner as in the frequency response analysis case. The unbalance
entry ID is same as that specified by RGYRO in case control.
$ Multiple Unbalance Loads
$ Rotor 1
UNBALNC 10 122.68 41 1. 0. 0.
1.-5 0.0 0.0 0.0 1000.
92 93 94
$ Rotor 2
UNBALNC 10 122.68 26 1. 0. 0.
1.-5 0.0 0.0 0.0 1000.
97 98 99
EPOINT 92 93 94 95
EPOINT 96 97 98 99
Results obtained for this model are in Figure 12-5, Figure 12-6 and Figure 12-7. The variation of rotor speed with
time is shown in Figure 12-5. For both the rotors, the rotor speed increases linearly with time till 0.5 seconds
and then it stays constant. Maximum speed for rotor 1 is 400 Hz, whereas for rotor 2, it is 800 Hz. The loads

Main Index
 234 MSC Nastran Rotordynamics User’s Guide
Introduction

due to unbalance for the rotor speed variation described earlier is shown in Figure 12-6. Here, the loads
increase with square of rotor speed till 0.5 sec and then it stays constant.

Figure 12-5 Variation of rotor speed with time

Figure 12-6 Unbalance Loads on the two rotors

The displacement produced due to unbalance loads is shown in Figure 12-7. For each of the rotor maximum
response is observed when the rotor speed is near the first forward whirl critical speed. For rotor 1, maximum
amplitude is observed at 0.375 sec where the rotor speed is around 290 Hz. Similarly, for the second rotor,
maximum amplitude is observed at 0.137 sec where the rotor speed is around 200 Hz. Both these frequencies
are in the neighborhood of their respective rotor`s first forward whirl critical speed.

Main Index
 Chapter 12: Multiple Rotors In Fixed System 235
Introduction

Figure 12-7 Variation of displacement and rotor speed with time

Examples used in this chapter are available under the following folder: ../tpl/ugrotor. A brief description of
these examples is given below:

Table 12-8
Input File Name Description
rug_12_1 Real Eigenvalue Analysis for Rotor 2
rug_12_2 Synchronous complex eigenvalue analysis for Rotor 2
rug_12_3 Synchronous complex eigenvalue analysis for model with 2 rotors
rug_12_4 Synchronous complex eigenvalue analysis for model with 2 rotors, at different
speeds
rug_12_5 Synchronous complex eigenvalue analysis for model with 2 rotors, at different
speeds with Rotor 2 as reference rotor
rug_12_6 Complex frequency analysis for model with 2 rotors
rug_12_7 Transient analysis for model with 2 rotors

Main Index
 236 MSC Nastran Rotordynamics User’s Guide
Introduction

Main Index
 Appendix A: Axisymmetric Harmonic Elements for Rotordynamic Analysis

Axisymmetric Harmonic
A Elements for Rotordynamic
Analysis


Introduction 238
 Element Details 238

Main Index
 238 MSC Nastran Rotordynamics User’s Guide
Introduction

Introduction
In the MSC Nastran 2013.1 release, axisymmetric harmonic elements were introduced for rotordynamic
analysis. For this purpose, two new axisymmetric harmonic elements are developed: CQUADX and
CTRIAX. These are similar to solid elements since the degrees of freedom are displacements (not rotations).
The CQUADX element has four vertex points and up to four (optional) mid edge points. The CTRIAX
element has three vertex points and up to three (optional) mid edge points. These elements resemble the
existing shell elements available in MSC Nastran (MSC Nastran Reference Manual, 2014) but refer to the
PAXSYMH card to define material properties.
A detailed description about dynamics of rotors and finite element discretization using axisymmetric
harmonic elements is provided in Geradin and Kill (1984) and Stephenson et. al (1989). In this appendix,
additional details related to implementation inside MSC Nastran are provided. The process used for
developing axisymmetric finite element models from an axisymmetric 3-D solid structure is shown in Finite
element modeling using axisymmetric harmonic elements.

Element Details

Figure 1-1 Finite element modeling using axisymmetric harmonic elements

Axisymmetric harmonic element analysis is performed in a cylindrical coordinate system. In the following,
“x” is the radial coordinate, “y” is the axial coordinate, and “θ” is the azimuthal angle in an element coordinate
system. Grid points lie in the (x,y) plane with “x” greater than or equal to zero, as shown in Finite element
modeling using axisymmetric harmonic elements. An error check is performed inside the code to ensure that
the axisymmetric rotor model being analyzed does not violate this requirement.
Let {r, y, θ}T represent the position vector for a grid point in the cylindrical coordinate system (r=radial,
y=axial, and θ=azimuthal). The displacements at a grid point using Fourier coefficients can be written as:

Main Index
 Appendix A: Axisymmetric Harmonic Elements for Rotordynamic Analysis 239
Element Details

where, “n” is the harmonic index and “θ” is the azimuthal angle. In this manner, non-axisymmetric solutions
may be found for axisymmetric structures. For linear analysis, the harmonics are not coupled, and the
harmonic analysis finds the solution for a single specified “n”. The value for “n” may be 0, 1, 2, 3, …. For
rotordynamic problems, n = 1 is used for lateral-bending analysis and n = 0 is used for axial/torsion analysis.
The solution variables are not local displacements, but coefficients of either or. For the Ur and Uy
displacements, the coefficients of represent solutions symmetric about the (x,y) plane while the coefficients
of represent solutions anti-symmetric about the (x,y) plane. For n = 0, and are used for expansion and is
used for torsion; is uncoupled from the other two.
For rotordynamic analysis, bending deformation that corresponds to n = 1 is sufficient to capture the
gyroscopic effects. For harmonic n = 1, the simplified equations are:

The strain displacement law for axisymmetric harmonic elements is given by:

 d 
 dr 0 0 
 
 0 d
0 
 ε rr   dy 
ε   
 rr   1 0
1 d  U
 r
ε   r dθ   
ε =  θθ  =  r  U y 
 ε rr   d d
0  U θ 
ε yy   dy dr 
   
ε θ r   0 1 d d 
 r dθ dr 
1 d d 1
 0 − 
 r dθ dr r 

This expression is used to determine the stiffness matrix for these elements.
The element displacement vector at each node can be written as:

For interpolation, standard Lagrangian shape functions [N] can be used. Using these shape functions, the
displacements can be written as:

Main Index
 240 MSC Nastran Rotordynamics User’s Guide
Element Details

where i = number of nodes in the element.

Based on the kinetic energy expression (Geradin and Kill, 1984), the element mass matrix is the following:

The gyroscopic matrix for the element is given by:

where

Each submatrix gi,j is the following:

gi,j =

The material properties may be isotropic or anisotropic, and they can also be temperature dependent. The
material temperature is assumed to be axisymmetric. If an anisotropic material is used, it is required that the
material be axisymmetric. Material damping is also supported in the analysis.

Constraints/Boundary Conditions
A configuration consisting of the new axisymmetric harmonic elements need not always define a rotating
structure. It could very well define a stationary structure like the casing of an engine or a water tank. A point
is defined to be on the axis of symmetry if its “X” coordinate is less than 1.E-4 from the axis of symmetry (see
Finite element modeling using axisymmetric harmonic elements). Points on the axis of symmetry require
special constraints to ensure continuity of displacements. These constraints are automatically supplied by the

Main Index
 Appendix A: Axisymmetric Harmonic Elements for Rotordynamic Analysis 241
Element Details

program. For externally applied forces and thermal loads, it is necessary to be consistent with the convention
used for matrices. The stiffness, mass and gyroscopic matrices are for 2π radians.
The new axisymmetric harmonic elements may share grid points with other spring and damper elements.
However, much care is needed. This connection is made between the harmonic displacement components of
the axisymmetric harmonic elements and the global displacement components of other elements. The usage
of the new connector element RBAX3D to make connections to a 3-D grid point is recommended.
The new RBAX3D connector helps to connect points of the axisymmetric harmonic elements to regular 3-
D points on the axis of symmetry. This has only been implemented for harmonic n=1and n=0.
Multi-point constraints may also be used with axisymmetric harmonic elements. Rigid body elements are
generally not compatible with these elements since rotation degrees of freedom are not used. However, they
can be used for the special case when there are two coincident grid points (no offset).

Main Index
 242 MSC Nastran Rotordynamics User’s Guide
Element Details

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D

User Defined Subroutines

B Using CBUSH2D

 Analysis in Windows Machine 244


Analysis on Linux 251

Main Index
 244 MSC Nastran Rotordynamics User’s Guide
Analysis in Windows Machine

Analysis in Windows Machine

This appendix describes steps involved in defining a user defined subroutine for usage in CBUSH2D entry.

Test machine Windows 7 Professional, DELL PRECISION M 4700

Requirements  Windows MSC Nastran ver_num i8 version
 SDK folder in: C:\MSC.Software\SDK\ver_num

Steps to modify the external CBUSH2D

1. Copy the implementation folder from the installation directory:
C:\MSC.Software\MSC_Nastran\ver_num\util_ver\nast\services\Implementations
(Available with standard MSC Nastran download)
To
E:\test_cbush2d
(You can put the implementation folder anywhere you prefer, if you do, modify the commands
appropriately)

2. Also you need to copy the three files SConopts,SConscript,SConstruct to the same folder as
E:\test_cbush2d
3. Put the buildit.bat into the same location as E:\test_cbush2d
4. Modify the buildit.bat so that the scons is pointing from:
C:\MSC.Software\SDK\ver_num\Tools\sconsi8
(or the correct directory on your computer)

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 245
Analysis in Windows Machine

The following is the contents of buildit.bat

----------------------------
setlocal
rem
rem batch file to build a SCA object
rem in order to build,
rem after unzipping the zip file.
rem *replace C:\SCA_Example\solver_SDK with the directory you put
the files in and run this file.
rem if desired, specify different locations for apps_local and
sca_object
rem
C:\MSC.Software\SDK\ver_num\Tools\sconsi8 %*
APPS_LOCAL=E:\test_cbush2d\sdk\apps_local
SCA_OBJECT=E:\test_cbush2d\sdk
Endlocal
--------------------------------
5. We can now modify the source code:
E:\test_cbush2d\Implementations\Elements\src\cbush2d\ext_cbush2d.F
Please note that the source file must be in the implementation folder.
6. We can build the object at E:\test_cbush2d
Open a DOS prompt at this location, and type: buildit cbush2d
Therefore we are only creating the object cbush2d at the APPS_LOCAL at
E:\test_cbush2d\sdk\apps_local\WIN8664I8\lib\SCA\MDSolver\Obj\Uds\Elements
You can see the cbush2d.dll there. Make sure that the cbush2d.dll has the correct time when you
execute the buildit script.

Main Index
 246 MSC Nastran Rotordynamics User’s Guide
Analysis in Windows Machine

7. In order to run MSC Nastran using your dll, it is recommend that you use a bat file to set the SCA
environment variables, and also point to the MSC Nastran exe correctly.
8. The following bat file is used: run_test.bat
set SCA_SERVICE_CATALOG=E:\test_cbush2d\sdk\apps_local\res\SCAServiceCatalog.xml
set SCA_RESOURCE_DIR=E:\test_cbush2d\sdk\apps_local\res
set SCA_LIBRARY_PATH=E:\test_cbush2d\sdk\apps_local\WIN8664I8\lib
set SCA_KERNEL_DEBUG=SvcLoad=all,ShrLibLoad=all,SvcTableDump=all,SvcCatalogDump=all
call nastver_num.exe beam_108_cbush2d old=no scr=yes notify=no
Name of the BAT file: run_test.bat. Make sure that your library path, and the MSC Nastran exe is at the
correct path.
In the DOS prompt you can type: run_test

Modifying ext_cbush2d.F
The information provided in the ELEMUDS is made available to ext_cbush2d.F through the following
arrays:
iarray : for integers
rarray : for real numbers
ciarray: for characters

Output produced by cbush2d routine are:

2x2 stiffness matrix (KXX, KYX, KXY and KYY)
2X2 damping matrix (CXX, CYX, CXY and CYX)

The input variable “freqva” corresponds to the frequency currently being used in the analysis. The code inside
ext_cbush2D.F can be customized to generate 2x2 stiffness and damping matrices for the current frequency
“freqva” using the inputs provided through ELEMUDS.

Example
Spring stiffness that needs to be included in the analysis:
Table 2-1
Frequency(f) KXX KYX KXY KYY
f=0 3.503e7 8.756e6 8.756e6 3.503e7
0 < f ≤ 45. 3.503e7 8.756e6 8.756e6 3.503e7
45 < f ≤ 85. 4.503e7 7.756e6 7.756e6 4.503e7
85 < f ≤ 115. 5.503e7 6.756e6 6.756e6 5.503e7
f > 115. 6.503e7 5.756e6 5.756e6 6.503e7

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 247
Analysis in Windows Machine

CXX=CYX=CXY=CYY=0
Changes required to input test deck and ext_cbush2d.F to obtain this distribution is described below:
a. Changes in input file:
File management statement:
CONNECT SERVICE TESTF 'SCA.MDSolver.Obj.Uds.Elements.cbush2d'

In Bulk data:
CBUSH2D 16 1000 911 912 XY
PBUSH2D 1000 0.0 0.0
ELEMUDS 1000 PBUSH2D TESTF cbush2d FREQ
real 0.0 3.503+7 8.756+6 8.756+6 3.503+7 0.00.0
0.0 0.0 45.4.503+7 7.756+6 7.756+6 4.503+7
0.00.00.00.0 85.5.503+7 6.756+6
6.756+6 5.503+7 0.00.00.00.0115.
6.503+7 5.756+6 5.756+6 6.503+7 0.00.00.0
0.0
b. ext_cbush2D.F
subroutine ext_cbush2d(freqva, iarray, rarray, ciarray, kxx, kyx,
& kxy, kyy, cxx, cyx, cxy, cyy, len_iarray, len_rarray,
& len_carray, elid, error_code)
implicit none

integer, intent(in) :: len_iarray, len_rarray, len_carray, elid

integer, intent(out) :: error_code
real, intent(in) :: freqva
real(8), intent(out) :: kxx, kyx, kxy, kyy, cxx, cyx, cxy, cyy

integer, intent(in), dimension(len_iarray) :: iarray

real, intent(in), dimension(len_rarray) :: rarray
integer, intent(in), dimension(len_carray) :: ciarray

character(len=4) :: carray
integer err
integer :: scaReturn
integer i
integer, external :: printf06
if (len_carray > 0) carray = transfer(ciarray, carray)
err = printf06("hello worldyy")
write(6,*) 'ext_cbush2d top: freqva', freqva
write(6,*) 'len_iarray, len_rarray',len_iarray,len_rarray
write(6,*) (rarray(i), i=1,len_rarray)

if (freqva.eq.rarray(1)) then
c
c nominal frequency call
c

Main Index
 248 MSC Nastran Rotordynamics User’s Guide
Analysis in Windows Machine

scaReturn = printf06('nominal call')

kxx = rarray(2)
kyx = rarray(3)
kxy = rarray(4)
kyy = rarray(5)
cxx = rarray(6)
cyx = rarray(7)
cxy = rarray(8)
cyy = rarray(9)
else if (freqva.le.rarray(10)) then
c frequency dep call
c
scaReturn = printf06('freq dep call')

kxx = rarray(2)
kyx = rarray(3)
kxy = rarray(4)
kyy = rarray(5)
cxx = rarray(6)
cyx = rarray(7)
cxy = rarray(8)
cyy = rarray(9)
else if (freqva.le.rarray(19)) then
kxx = rarray(11)
kyx = rarray(12)
kxy = rarray(13)
kyy = rarray(14)
cxx = rarray(15)
cyx = rarray(16)
cxy = rarray(17)
cyy = rarray(18)
else if (freqva.le.rarray(28)) then
kxx = rarray(20)
kyx = rarray(21)
kxy = rarray(22)
kyy = rarray(23)
cxx = rarray(24)
cyx = rarray(25)
cxy = rarray(26)
cyy = rarray(27)
else if (freqva.gt.rarray(28)) then
kxx = rarray(29)
kyx = rarray(30)
kxy = rarray(31)
kyy = rarray(32)
cxx = rarray(33)
cyx = rarray(34)
cxy = rarray(35)

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 249
Analysis in Windows Machine

cyy = rarray(36)
endif

write(6,*) 'inside EXT_CBUSH2D'

write(6,*) 'Dimensional Stiffness (lb/in):(KXX,KYX,KXY,KYY)'
write(6,'(ES10.3E1)')KXX,KYX,KXY,KYY
write(6,*) 'Dimensional Damping (lb*s/in):(CXX,CYX,CXY,CYY)'
write(6,'(ES10.3E1)') CXX,CYX,CXY,CYY

end subroutine ext_cbush2d

Example
In order to verify if the code works as expected, analysis is performed using SOL 107 and SOL 108 and results
obtained are compared with those obtained for a test deck using DMIG entries.
SOL 107
For the SOL 107 analysis, the spring and damping stiffness corresponding to f = 0 Hz are used in the analysis.
Corresponding DMIG entries in beam_107_sync_dmig.dat
DMIG STIF 0 1 1
DMIG STIF 91 1 91 1 3.503+7
91 2 8.756+6
DMIG STIF 91 2 91 1 8.756+6
91 2 3.503+7
Test deck with CBUSH2D used for comparison is: beam_107_sync_cbush2d.dat

Identical results are obtained for critical frequencies using both the decks, as shown below.

SOL 108
Frequency range used for the analysis:
FREQ1 100 30.0 10.0 10
In order to compare the results, four different test cases are used with varying spring stiffnesses set up

Main Index
 250 MSC Nastran Rotordynamics User’s Guide
Analysis in Windows Machine

using DMIG entries.

Table 2-2
Test Deck KXX KYX KXY KYY
beam_108_dmig1.dat 3.503e7 8.756e6 8.756e6 3.503e7
beam_108_dmig2.dat 4.503e7 7.756e6 7.756e6 4.503e7
beam_108_dmig3.dat 5.503e7 6.756e6 6.756e6 5.503e7
beam_108_dmig4.dat 6.503e7 5.756e6 5.756e6 6.503e7

a. Comparison with beam_108_dmig1.f06

Here, results match for 30 and 40 Hz frequencies.

b. Comparison with beam_108_dmig2.f06
Here, results match for 50, 60, 70 and 80 Hz frequencies.

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 251
Analysis on Linux

c. Comparison with beam_108_dmig3.f06

Here, results match for 90, 100 and 110 Hz frequencies.

d. Comparison with beam_108_dmig4.f06

Here, results match for 120 and 130 Hz frequencies.

Analysis on Linux
Requirements:
1. MSC Nastran ver_num i8 version
2. SDK installation is separate and must be installed. In this example as shown below it is installed in
/scratch/ssp/sdk_ver_num. Make a note of the Tools Folder. The sconsi8 executable resides here.

Main Index
 252 MSC Nastran Rotordynamics User’s Guide
Analysis on Linux

Compiling External CBUSH2D

1. Copy the services folder from the MSC Nastran installation directory to a local directory.
This example has copied them to /scratch/ssp/services.
Following shows the MSC Nastran installation where services directory can be found.

Only files pertaining to CBUSH2D under the elements subdirectories need to be present. Others can be
deleted. This will improve the compilation time.
Your directory structure should look similar to images shown below.

Your directory structure should look similar to images shown below.

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 253
Analysis on Linux

2. Change over to src directory where the routine is located. (shown in image below.)

3. We can now modify the source code:ext_cbush2d.F

Please leave the source file in this directory. Scons will automatically recognize it and compile/link it
when scons is run in step5.
4. After modifying the routine switch the directory to root level where services are installed. In this
example /scratch/ssp/services is being used.
5. To compile and link the routine to MSC Nastran issue the scons command

For the i8 version: /scratch/ssp/sdk_ver_numTools/sconsi8 > scons.log

This step will go thru and creation of Object and Apps local directories and end with
scons: done building targets.

Main Index
 254 MSC Nastran Rotordynamics User’s Guide
Analysis on Linux

After the successful build in the services folder other directories would have been created and folder will look
as follows.

Environment variables to this path must be setup prior to the running MSC Nastran.
For mode = i8, libcbush2d.so is available in:
/scratch/ssp/services/ LX8664I8_RHE63/Implementations/Elements/src/cbush2d
6. Following SCA environment variables needs to be set prior to running the MSC Nastran. Depending
on the shell being used following environment variables should be set.
Following is the example using csh.
setenv SCA_SERVICE_CATALOG
/scratch/ssp/services/Apps/res/SCAServiceCatalog.xml
setenv SCA_RESOURCE_DIR /scratch/ssp/services/Apps/res
setenv SCA_LIBRARY_PATH /scratch/ssp/services
/Apps/LX8664I8_RHE63/lib

7. After setting the environmental variables MSC Nastran job can be submitted using following
command.
prod_ver test_cbush2d old=no scr=yes notify=no mode=i8

Definitions Used in this document

The Rotordynamics guide uses certain definitions to denote installation directories, and product versions of
MSC Nastran.

install_dir The full path to the directory used in the installation

Example: The system RC file is install_dir/conf/nast2018rc.
prod_ver The Product and Version of MSC Nastran
Example: For MSC Nastran 2018 prod_ver=nast2018
ver_num The version number. For MSC Nastran 2018, this is 2018.
util_ver The version number used for building utilities.

Main Index
 Appendix B: User Defined Subroutines Using CBUSH2D 255
Analysis on Linux

Example: For MSC Nastran 2018 util_ver=msc20180

msc_arch The architecture of the platform.
Example: win64i8 for Windows 64 i8 version
Example: linux64i8 for Linux 64 i8 version

Main Index
 256 MSC Nastran Rotordynamics User’s Guide
Analysis on Linux

Main Index
 Nomenclature
MSC Nastran Implicit Nonlinear (SOL 600) User’s GuideRotordynamics User’s Guide

Nomenclature

Main Index
 258 MSC Nastran Rotordynamics User’s Guide

[B] Damping in the model

[BS] Damping due to non-rotating portion of the model (Stator Damping)
[BR] Damping due to rotating portion of the model (Rotor Damping)
[BHR] Hybrid damping for the rotor
[BHS] Hybrid damping for the stator

Circulation matrix due to [BR]

 BRC 

Circulation matrix due to [BHR]

 BH RC 

[DR] Viscous damping matrix in the rotating system

{FR} External force vector in rotating coordinate system.

{FS} External force vector in fixed coordinate system
[K] Stiffness matrix for the model
[Kc]R Circulation terms due to damping in rotating components
[Kc]S Circulation terms due to damping in fixed components
[KS] Stiffness of the non-rotating portion of the model
[KG] Differential Stiffness due to axial centrifugal load
[KHS] Hybrid structural damping for stator
[K4R] Material damping for the rotor part
[KHR] Hybrid damping for the rotor part

α1[ M RC ] + α 2 [ K RC ] Circulation matrix due to Rayleigh damping for rotors

GR Circulation matrix due to structural damping for rotors

 K RC 
WR 3 

1 Circulation matrix due to structural damping for rotors

 K 4 CR 
WR 4 
[M] Total mass matrix for the model
[MS] Mass matrix for the stator portion

Main Index
 Nomenclature 259

[MR] Mass matrix for the rotor

[G] Gyroscopic matrix for the rotor
[T] Transformation matrix from fixed to rotating coordinate system
[Z] Centrifugal softening matrix
D Dissipative energy
G Structural damping parameters for stator
GR Structural damping parameters for rotor
g Damping Coefficient
g(ω) Complex frequency response
i –1

{ga(p)} Modeshape corresponding to complex eigenvalue p

L Lagrange`s Potential
Id, Ip Diametral and polar moment of inertia
N(t) Nonlinear forces due to bearing elements
p Complex eigenvalue
U System`s Potential Energy
W Work done by non-dissipative forces
W3, W4, WH Parameters for converting structural damping to equivalent viscous damping for
stator
WR3, WR4, WRH Parameters for converting structural damping to equivalent viscous damping for
rotors
u,v Translational velocities in the fixed system in X and Y direction
α, β Parameters determined by a least-mean-square fit of the relative rotor speeds
input on the RSPINR between SPDLOW and SPDHIGH of the reference rotor
α1, α2 Rayleigh Damping Parameters for stator
α1R, α2R Rayleigh Damping Parameters for rotor
Ω Rotor Operating Speed

Rotation and angular velocity about Z axis in the fixed frame

φ , φ

θ ,θ Rotation and angular velocity about X axis in the fixed frame

ψ ,ψ Rotation and angular velocity about Y axis in the fixed frame

Main Index
 260 MSC Nastran Rotordynamics User’s Guide

The instantaneous angular velocities about the X , Y and Z (Rotating

ω x, ω y, ω z coordinate system)

Derivative with respect to time

Variable defined in rotating coordinate system

Main Index
 Glossary

Glossary

Main Index
 262 MSC Nastran Rotordynamics User’s Guide

Asynchronous Analysis Analysis where the rotor speed is specified by the user and independent of
frequency of excitation of eigenfrequency.
Axisymmetric harmonic Elements defined using Fourier coefficients for rotordynamic analysis, mostly
elements analyzed with harmonic index = 1 for lateral bending.
Campbell Diagram Plot that shows the variation of whirl frequencies with rotor speed.
Centrifugal Softening Reduction in bending stiffness observed when the problem is solved in a
rotating reference frame. It is proportional to the square of rotor speed.
Circulatory matrix Skew symmetric matrix proportional to rotor speed generated when the
damping defined in a rotating reference frame is transferred to a fixed reference
frame and vice versa, added to stiffness matrix [K].
Coriolis Matrix Skew symmetric matrix proportional to rotational speed for analysis in a
rotating reference frame, added to damping matrix [B].
Critical Speed Rotational speeds at which vibration due to rotor unbalance is a local
maximum.

It can also be defined as a rotational speed of the system when a multiple of

rotor speed coincides with one of the system's natural frequencies.
Differential/Geometric Additional stiffness due to axial centripetal force in a rotating reference frame.
Stiffness
External Damping Damping associated with the stator when problem is analyzed in fixed
reference frame.
Gyroscopic Matrix Skew symmetric matrix proportional to rotational speed for analysis in a fixed
reference frame, added to damping matrix [B].
Internal Damping Damping associated with the rotor when problem is analyzed in fixed reference
frame.
Mode Tracking For Campbell diagram plotting, it is useful to track rotor modes in case the
eigenvalues of the modes change order.
Reference Rotor Rotor specified in the RGYRO entry, in case multiple rotors are present in the
analysis.
Rotor Unbalance When the rotor mass centerline does not coincide with its rotational axis, then
mass unbalanced inertia related rotating forces occur.

Main Index
 Glossary 263

Rotor Whirl Planar motion of a rotor is called a whirling motion or a whirl.

Backward Whirl: Here, the rotor whirl direction is opposite to that of the
applied angular velocity.

Forward Whirl: Here, the rotor assembly whirls in the same direction as the
applied angular velocity.
Synchronous Analysis Analysis where rotor speed is same as the actuation frequency (for frequency
response analysis) or same as the eigenfrequency (for complex eigenvalue
analysis)

Main Index
 264 MSC Nastran Rotordynamics User’s Guide

Main Index
 References
MSC Nastran Implicit Nonlinear (SOL 600) User’s GuideMSC Nastran DMAP Program-
mer’s Guide

References

Main Index
 266 MSC Nastran Rotordynamics User’s Guide

Adams, M. L., Padovan, J., & Fertis, D. G. (1982). Engine dynamic analysis with general nonlinear finite-
element codes, Part 1: Overall approach and development of bearing damper element. Journal of Engineering
for Gas Turbines and Power, 104(3), 586-593.
Branagan, L. A., Barrett, L. E., & Cloud, C. H. (). A manual for use with tilting pad bearing program
THPAD. University of Virginia, Romac Report, (284).
Bellinger, D. (1985), Dynamic Analysis by the Fourier Transform Method with MSC.Nastran, Proceedings
1995 MSC World Users' Conference.
Dimarogonas, Andrew D., Stefanos A. Paipetis, and Thomas G. Chondros. (2013) Analytical methods in rotor
dynamics. Springer Science & Business Media.
Friswell, M. I., Penny, J. E. T., Garvey, S. D. and Lees, A. W. (2010). Dynamics of rotating machines.
Cambridge University Press.
Genta, Giancarlo (2007). Dynamics of rotating systems. Springer Science & Business Media.
Geradin M. and Kill N., (1984), “A new Approach to Finite Element Modeling of Flexible Rotors. “
Engineering Computations, Vol. 1, 52-64.
Harris, C. M. (1987), Shock and Vibration Handbook, Page 4-8, 3rd Edition, McGraw Hill
Jeffcott, H. H. (1919). XXVII. The lateral vibration of loaded shafts in the neighbourhood of a whirling
speed.—The effect of want of balance. The London, Edinburgh, and Dublin Philosophical Magazine and
Journal of Science,37(219), 304-314.
Kumar, M. Santhosh (2011). “Rotor dynamic analysis using ANSYS.” IUTAM Symposium on Emerging
Trends in Rotor Dynamics. Springer Netherlands.
MSC Nastran Dynamic Analysis Users Guide, Version 2017 (2017), MSC Software, Newport Beach, Los
Angeles, CA.
MSC Nastran Quick Reference Guide, Version 2017 (2017), MSC Software, Newport Beach, Los Angeles, CA.
MSC Nastran Reference Manual, Version 2017 (2017), MSC Software, Newport Beach, Los Angeles, CA.
MSC Nastran Superelements User’s Guide, Version 2017 (2017), MSC Software, Newport Beach, Los Angeles,
CA.
MSC Nastran User Defined Services User`s Guide, Version 2017 (2017), MSC Software, Newport Beach, Los
Angeles, CA.
MSC Nastran SCA Service Guide, Version 2017 (2017), MSC Software, Newport Beach, Los Angeles, CA.
Nandi, A., and Neogy S., (2001), “Modelling of rotors with three-dimensional solid finite elements.” The
Journal of Strain Analysis for Engineering Design 36(4), 359-371.
Nelson, F. C. (2007). Rotor dynamics without equations. International Journal of COMADEM, 10(3), 2.
Nelson, H. D. and Vaugh, J. M. (1976), “The Dynamics of Rotor-Bearing Systems Using Finite Elements”,
Journal of Manufacturing Science and Engineering 98(2), 593-600.
Rao, J. S. (2011) History of rotating machinery dynamics. Vol. 20. Springer Science & Business Media.
Seshu, P. (1997) “Substructuring and component mode synthesis.” Shock and Vibration 4.3 Pages: 199-210.

Main Index
 References 267

Shanmugam A., Padmanabhan C., (2006), “A fixed–free interface component mode synthesis method for
rotordynamic analysis”, Journal of Sound and Vibration, Volume 297, Issues 3–5, Pages 664-679.
Stephenson R. W., Rouch K. E. and Arora R., (1989), “Modelling of Rotors with Axisymmetric Solid
Harmonic Elements” Journal of Sound and Vibration 131(3), 431-443.
Ushida, A. and Chua, L.O. (1984), Frequency-domain analysis of nonlinear circuits driven by multi-tone
signals, IEEE Trans. Circuits Syst., Vol. CAS-31, pp. 766-778
Vollan, Arne, and Louis Komzsik (2012). Computational techniques of rotor dynamics with the finite element
method. CRC Press.
Vance, John M., Fouad Y. Zeidan, and Brian Murphy. (2010) Machinery vibration and rotordynamics. John
Wiley & Sons.
Vollan, Arne, and Komzsik, L. (2012). Computational Techniques of Rotor Dynamics with the Finite Element
Method. CRC Press.
Wagner M. B., Younan A., Allaire P., and Cogill R., (2010) “Model Reduction Methods for Rotor Dynamic
Analysis: A Survey and Review,” International Journal of Rotating Machinery, vol. 2010, 17 pages.
Zu, J.W. and Ji, Z.Y. (1998) "Steady-State Response of Continuous Nonlinear Rotor-Bearing Systems Using
Analytical Approach", Journal of Engineering for Gas Turbines and Power, ASME, 120, pp 751 - 758.

Main Index
 268 MSC Nastran Rotordynamics User’s Guide

Main Index