You are on page 1of 24

# 5/26/2011

## CAESAR II: Calculating Modes of Vibration

A Quick Overview
26 May 2011
Presented by David Diehl

Quick Agenda

## Modal Extraction, a brief introduction

Dynamic Input Review
Results Review
Use as Acceptance Criteria
Close

1
5/26/2011

INTRODUCTION
Modal Extraction / Eigen Solution

## M &x& + C x& + Kx = F ( t ) is the angular frequency

C =0 oscillation
F (t ) be harmonic
There is a matching shape to this
so oscillation
x = A sin t
&x& = 2 A sin t = 2 x There is no magnitude to this shape
2 Mx + Kx = F ( t )
This is important:
let Think of a mode of vibration (the &
F (t ) = 0 mode shape pair) as a single degree
(K M 2 )x = 0 of freedom system
so
x =0
or
K M 2 = 0
= K M

2
5/26/2011

## Two examples of a One

Degree of Freedom
(DOF) System A two DOF System An n DOF System

Mode 2

Mode 3

Mode 4

Mode n

## We are following nodal displacement distortion of the pipe centerline

The pipe also has modes of vibration associated with shell distortion:

## :From Piping Vibration Analysis

by J.C. Wachel,
Scott J. Morton and
Kenneth E. Atkins of
Engineering Dynamics, Incorporated
San Antonio, TX

## A Tutorial from the

Proceedings of 19th Turbomachinery Symposium

3
5/26/2011

## DYNAMIC INPUT REVIEW

Controlling the Analysis

4
5/26/2011

5
5/26/2011

## General Comments on Data Entry

line below Delete
current selected
line(s)

Save,
Error Check
Check,
Run

Comment
(do not process)

Modifying Mass

6
5/26/2011

Modifying Mass

X, Y, Z or ALL
The affected
Or a range
TheA signed
zero or Node number
of Nodes
magnitude
eliminates
li i t the is
th RX, RY, RZ or RALL
summedmass. the
with
calculated mass.

Calculated Mass:
Node Node Node

## Remember, damping was

eliminated from the equation of
motion (C=0). Point damping
is simulated with a stiff spring.
Mechanical Hydraulic

7
5/26/2011

Control Parameters

Def=Default;
this is a button

Entry cell
(use F1 for help)

Nonlinear Considerations

Our equation of motion insists on a linear system that is, the stiffness, K, is
constant. ( K M 2 ) x = 0

## The dynamic model must linearize those nonlinear conditions.

In many cases, the operating state of nonlinear boundary conditions can serve
as the linear state for the dynamic evaluation.

8
5/26/2011

: Cold Position

A +Y
(resting)
restraint

Liftoff

Dynamic Model
(no restraint)

9
5/26/2011

## : (Static) Operating Position 2

No liftoff

Dynamic Model
(double-acting Y)

## X : (Static) Operating Position

Friction defined;

Dynamic Model

X K
K
Z

10
5/26/2011

## This Stiffness Factor for Friction is not a physical parameter; it is a modeling

tool.

Larger
g normal loads ((N)) will p
produce g
greater restraint

This is NOT a 0 or 1! I use 1000 but values as low as 200 produce similar
results for the models I run.

This value will knock out frequencies associated with frictionless surfaces.

ASCE 7-10 para. 15.5.2.1: "Friction resulting from gravity loads shall not be
considered to provide resistance to seismic forces
(But were
we re not running a seismic analysis here
here.))

11
5/26/2011

## Are all mode important?

No,
N nott for
f our purposes.

The lower (frequency) modes contribute the greatest structural response of the
system.

CAESAR II extracts modes starting with the lowest mode (lowest frequency).

## Piping modes of vibration above 33 Hertz do not show resonant response to

seismic motion. This is the default CAESAR II cutoff frequency.

Piping modes of higher frequency (100+ Hz) may play a role in fast-acting
events such as fluid hammer.

## Piping modes at lower frequencies respond to many environmental harmonic

loads (equipment vibration, acoustic vibration & pulsation).

## Two parameters are checked to stop the eigensolution:

A maximum frequency.

12
5/26/2011

## Lumped Mass versus Consistent Mass

For many years CAESAR II (like most analysis tools) ignored rotational inertia
and off-diagonal mass terms.

## This is what we call lumped

p mass.

Todays bigger and faster PCs can handle the fully-developed, complete mass
matrix.

## Consistent mass will more accurately determine the frequencies of natural

vibration without adding more nodes (mass points) to the static model.

BUT more mass points may still be required to establish a proper mode
shape in the frequency/mode shape pair.

13
5/26/2011

14
5/26/2011

## The Sturm sequence check is a back check on the calculated frequencies

View the eigensolver as a search routine that finds system natural frequencies
from lowest to highest.
g

## The Sturm sequence check as a separate calculation of the total number of

modes below the last frequency produced. If this count doesnt match the
eigensolver total, the program will state that the check has failed.

## Control Parameters (confirming the modal solution)

15
5/26/2011

RESULTS REVIEW
What Does It All Mean?

16
5/26/2011

## Results Frequency Report

f t

secondper second
seconds per cycle

## Mode shapes (mass & unity normalized)

Modes Mass Normalized the tendency of that modes contribution to the overall
response to a quickly-applied load, all other things being equal (i.e. DLF and point
application).

Model Unity Normalized the typical mode shape. This is the same shape but
normalized to one.

Same shape;
different magnitude

17
5/26/2011

## Results Mass Model

: Lumped Mass

Consistent Mass :

## Results Active Boundary Conditions

Input
Operating Position (Liftoff 30, Resting 40)

18
5/26/2011

Results Animation

Is the Static Model Sufficient?

19
5/26/2011

## Is the static model adequate?

More mass points may be required to approximate the continuous mass beam

Reality:

CAESAR II:

## half of total mass at end

10 20

g more nodes improves
p the calculation

## Is the static model adequate?

hand
2node 2node 10node 10node 100node
Mode calculation
lumped consistent lumped consistent lumped
(continuous)

## 1 0.328 0.473 0.469 0.479 0.471 0.471

2 1.51 2.902 2.971 2.948 2.95
3 4.658 8.039 8.235 8.248 8.26
4 57.339 15.572 16.005 16.143
5 25.415 26.377 26.646

## OD=4.5 in Consistent mass will develop

t=0.237 in better frequencies
***BUT***
length=50 ft
density=0.283 lb/cu.in More mass points may be
E=29.5E6 psi needed to develop the mode
shapes

20
5/26/2011

## Some simple suggestions:

Add nodes (break pipe) so that the maximum node spacing is no more
than one foot ((300mm)) p
per nominal inch of p
pipe
p

## from the paper On Mass-Lumping Technique for Seismic Analysis of

Piping - John K
Piping K. Lin & Adolph T
T. Molin of United Engineers &
Constructors and Eric N. Liao of Stone & Webster

L = 4 9.2( D 3 t W )

An End in Itself

21
5/26/2011

## Use as an acceptance criteria

The lowest natural frequency can be used to assess the risk of failure
associated with dynamic response

## Adding stiffness may impact thermal flexibility

CLOSE

22
5/26/2011

Closing Points

Many systems are built for static loads (deadweight and thermal strain) by
providing Y supports alone, leaving great flexibility in the horizontal plane
modal analysis will uncover such oversights.

system response.

PDH Certificate

23
5/26/2011

## Intergraph @ Hexagon 2011

www.hexagonconference.com/ppm