An FEA-Based Acoustic Fatigue Analysis

Methodology
Timothy C. Allison, Ph.D.
Lawrence J. Goland, P.E.
Southwest Research Institute
San Antonio, TX
ANSYS Regional Conference:
Engineering the System
August 31 - September 1, 2011
Houston, TX

Outline
Introduction and Theory
Existing Acoustic Fatigue/AIV Screening
Methods
Carucci-Mueller, Eisinger
Energy Institute

SwRI Method
AIV Solutions
Conclusions

Introduction
Acoustically Induced Vibration (AIV) refers
to high-frequency vibration (typically 5001500+ Hz) in piping downstream of a
pressure-reducing device
E.g. a control valve or pressure relief valve

Can result in high cycle fatigue failures,

particularly at branch connections
First identified in 1983 by Carucci and
Mueller
Often a concern in flare/blowdown piping
with thin walls and large diameters.

Image Courtesy Tyco Valves & Controls

Theory Overview
AIV is caused by the following four physical
phenomena:
Excitation from a pressure-reducing valve causes highfrequency pressure fluctuations in downstream piping.
These fluctuations excite higher order acoustic modes in
the pipe with circumferentially varying pressure mode
shapes.
The acoustic pulsations couple to shell modes of the
main piping.
Branch connections or other welded discontinuities in
the main line serve as stress risers.

Theory: Acoustic Cross Modes

Theory: Pipe Shell Modes

Existing AIV Screening Methods

Carruci-Mueller paper (1983) introduced design
limits based on failure/non-failure experience.
PWL and Pipe Diameter

Eisinger (1997) modified the Carruci-Mueller

limits to include different excitation parameter
and wall thickness.
M*P and Pipe D/t Ratio

Eisinger later (1999) used FEA to extend the

design curve to lower D/t ratios.

Existing AIV Screening Methods (2)

Eisinger Design Curve

Carruci-Mueller Design Curve

Existing AIV Screening Methods (3)

The Energy Institute (2005) introduced a screening
methodology for AIV:
Simple source PWL computation
PWL decay to branch connection and addition of PWL
from multiple sources at each branch
Estimate of fatigue life from curve-fit data (data from FE
models calibrated to historical failure/non-failure data)
Fatigue life estimation including reduction due to
weldolet fittings and small branch diameter to main line
diameter ratios
Likelihood of Failure (LOF) computed from estimated
fatigue life

SwRI Method Overview

Valve excitation analysis, acoustic analysis and
finite element analysis performed to determine
coincidence of acoustic and pipe shell modes
Forced response analysis of FE model at coincident
modes performed with shell models to determine
stresses at fillet weld and resulting fatigue life.
Excitation from valve amplified by acoustic amplification
factor to account for acoustic resonance
Stresses evaluated using mesh-insensitive procedure for
welded joints in accordance with Section 5.5.5 of the
ASME Boiler and Pressure Vessel Code, Section VIII,
Division 2

SwRI Method Valve Excitation

Valve excitation analysis performed
using control valve noise prediction
standard IEC 60534-8-3
Detailed source PWL prediction
Peak noise frequency from vena
contracta velocity and jet diameter

Model PWL decay to branch and

summation of sources at branch in
same manner as Energy Institute
method
Convert PWL to SPL and dynamic
pressure

Image Courtesy Floyd Jury, Fisher Controls

SwRI Method Acoustic Modes

Closed form solution used to
model higher-order acoustic
modes
Resulting acoustic
frequencies and mode
shapes validated with ANSYS
Acoustic 3D FEA models
Multiply valve excitation by
amplification factor to
account for acoustic
resonance

p1

p2
p3

q1

p4
p5

p6

SwRI Method Pipe Shell Modes

ANSYS APDL scripts constructed to efficiently
construct shell element models of piping at
branch connections
Modal analysis performed for each connection
over excitation frequency range
Results postprocessed externally via spatial FFT
to determine dominant nodal diameter patterns
in each mode

SwRI Method Pipe Shell Modes (2)

SwRI Method Pipe Shell Modes (3)

Circumferential Mode
Shape (n)

FFT performed in order to

identify n

SwRI Method Coincidence

SwRI Method Forced Response

SwRI Method Forced Response (2)

Mesh-Sensitive Peak Stress

Coincident
Mode

1150

1170

1190

1210
1230
Frequency, Hz

1250

1270

1290

SwRI Method Forced Response (3)

Note: Stresses shown are mesh-sensitive and are not accurate

absolute values. Mesh-insensitive stresses are calculated with
ASME B&PV Code Sec VIII Div 2 procedure

SwRI Method Forced Response (4)

At Fillet
Weld
Toe

Note: Stresses shown are mesh-sensitive and are not accurate

absolute values. Mesh-insensitive stresses are calculated with
ASME B&PV Code Sec VIII Div 2 procedure

SwRI Method Fatigue Life

Use relative stress
distribution from meshsensitive results to find
location of maximum
stress
Use nodal forces and
moments to calculate
bending and membrane
stresses and assess
fatigue life
Images Courtesy ASME Boiler & Pressure Vessel Code, Section VIII, Division 2

SwRI Method Fatigue Life

ASME Div 2 master S-N developed based on a large
amount of welded pipe and plate joint fatigue test data
Fatigue life assessed on -3 S-N curve for <1%
probability of failure

Image Courtesy Dong 2009

Conclusions
New AIV analysis methodology developed based
on physical principles
Method uses automated implementation of
valve noise prediction standard and exact
acoustic solution for efficient excitation solution
Automated scripting tools applied for efficient
FEA solution of coincident stress at connection
and mesh-independent fatigue life results
FEA-based approach allows for modeling of
various countermeasures

Method Comparison
CarruciMueller

Eisinger

Energy
Institute

SwRI
Method

Calculates PWL

Includes Pipe Diameter

Uses historical data

See (1)

Includes pipe wall thickness

Includes multiple sources & decay

Includes connection type

Includes branch diameter

Includes acoustic/structural coincidence

Includes excitation frequency

Allows detailed analysis of design

alternatives

Fatigue Life Calculation

1Future

See (2)

work includes validation of method with test and historical data

2Calculated

fatigue life is part of calibrated screening procedure, not end result

QUESTIONS?