Documente Academic
Documente Profesional
Documente Cultură
PVP2014
July 20-24, 2014, Anaheim, California, USA
PVP2014-28260
STRAIN BASED AIV EVALUATION
J. Adin Mann III
Emerson Process Management
Marshalltown, Iowa, USA
Allen C. Fagerlund
Emerson Process Management
Marshalltown, Iowa, USA
ABSTRACT
A framework for Acoustic Induced Vibration (AIV)
evaluation is outlined which is based on estimating the pipe
surface strain for an evaluation framework of structural fatigue.
Critical to this approach is that the assessment is implemented
with frequency based formulations. The frequency based
formulation allows for more accurate determination of the
pipes structure response and combining different sources, such
as the valve and piping elements. The approach relies on
internationally recognized standards as the core technology, in
particular the IEC 60534-8-3 control valve aerodynamic noise
prediction standard and the fatigue assessment in design codes,
such as the ASME Boiler and Pressure Vessel code. These are
augmented with system noise predictions using a nondimensional testing based model for piping component noise
predictions. Components of this approach have been described
in previous papers and are presented here in a more complete
form.
INTRODUCTION
Acoustic Induced Vibration (AIV) can cause catastrophic
failures in piping systems downstream of noise generating
segments of the piping.1,2 Noise generating segments include
valves, expansions, elbows, and tees. Any assessment scheme
needs to (a) predict the noise internal to the pipe, (b) predict the
impact of the noise on the pipe wall, and (c) compensate for
fabrication and reinforcement methods.
A criteria based on a maximum sound level has been used3.
One of the most widely used approaches to assessing the
potential impact of AIV has been to use a set of criteria curves
based on an approximate noise prediction equation, commonly
called the Carucci-Mueller method.1,4,5,6 In this approach, the
impact of the noise on the pipe wall is estimated by a single
curve which has been developed from evidence of failures in
field data. This method has been considered sufficiently
conservative for many applications, but concerns have been
raised about the extension of the method to pipe sizes for which
failure data is not available. The design curve for pipe sizes
that are outside of the available data have been addressed by
several people with their rational for how the failure curve
Daniel Eilers
Emerson Process Management
Marshalltown, Iowa, USA
NOMENCLATURE
a
Pipe internal radius
Internal pipe diameter
Di
Center frequency of frequency band i
fi
FSRF Fatigue Strength Reduction Factor
Sound level decay for source n
LDrn
Lpe,1m IEC external sound pressure 1m from pipe surface
External sound pressure on the pipe wall surface
Lpe,S
Total external sound level with all sources and decay
LpeT
LpeT_REV Total external sound level revised for fatigue
strength reduction and stress concentrations factors
Internal sound pressure generated by valve trim
Lpi
Internal sound pressure generated valve outlet
LpiR
expansion
Valve noise inside the pipe
LpiS
Molecular weight
MW
P
Pressure drop across valve
Valve inlet pressure
P1
Reference sound pressure (20x10-6 Pa)
pref
External sound pressure on the pipe wall surface
psurf
PWL Carucci-Mueller Internal Sound Power Level
r
Distance downstream of the noise source
SCF
Stress concentration factor
Sp
Spectral density from a piping element
SPLREV Sound level revised to include fatigue strength
reduction and stress concentration factors
T
Temperature
TL
Transmission Loss
Pipe wall thickness
ts
U
Average velocity
W
Mass flow rate
f
Frequency band range
Non-dimensional frequency
P 3.6 T
W 2
PWL = 10 log10
P1
MW
1.2
+ 126.1 (1)
D + 2t s + 2
(2)
L pe,1m ( f i ) = L piS ( f i ) + TL( f i ) 10 log10 i
Di + 2t s
but has separated the formulation into three components:
TL( f i )
the valve
transmission of the internal sound pressure to
the sound pressure on the outside surface of
the pipe
D + 2t s + 2
radiation from the outside pipe
10 log i
Di + 2t s
surface to 1m from the pipe wall
Thus, the formulation allows one to use the components as
needed. For the application of AIV, only the first two terms
will be used to predict the sound pressure on the surface of the
pipe outside wall:
(3)
L pe ,s ( f i ) = L piS ( f i ) + TL( f i )
The formulation also breaks the internal sound level from
the valve into two components:
L piS ( f i ) = 10 log10 10
L pi / 10
+ 10
L piR / 10
(4)
where
TRANSMISSION LOSS
One of the central points of the investigations to extend the
Carucci-Mueller approach to new facility designs has been
extending the design curve to larger and thinner pipes as well as
smaller pipes. The key issue has been determining the
dependence of the pipe wall vibration on the ratio of the
diameter and wall thickness. Various theories have been applied
to the criteria, resulting in different criteria curves. The IEC
prediction methodology, however, presents an opportunity to
use the transmission loss theory that takes into account the
frequency based behavior of how the internal sound couples to
the vibration in the pipe wall. In different frequency ranges, the
transmission loss has a different dependence on the diameter
and wall thickness. Thus, the appropriate model depends on the
dominant frequency range of the noise generated within the
pipe. This depends on the pipe dimensions, but also the
characteristics of the pipe. Thus by using the IEC transmission
loss model the appropriate dependence on diameter and wall
thickness is automatically included, and the impact on pipe wall
vibration depends on the frequency characteristics of the
internal noise.
The data13 in Figure 1 were taken at the Fisher and Gulde
laboratories as well as from the literature14 on a variety of pipes
near the first internal coincidence frequency. This is only for
air as the internal and external fluid under static flow
conditions, however, it indicates the concept that there are
ranges where (t/D) and (t/D)2 are appropriate scaling
parameters. The approximate dividing point is at (t/D)=.03 for
an air-steel-air system. All of the pipes in the Carucci&Muller
study are below this point and track with (t/d), while the pipes
in the Chiyoda study8 are above this and (t/d)2 is preferred.
TL ~ (t/D)
75
70
65
Transmission Loss
make design and cost tradeoffs between valves with low outlet
noise compared to fabrication of the downstream piping.
A significant improvement with the current IEC noise
prediction is gained because the formulation is frequency
based. Each term is predicted in frequency bands and then the
results added to obtain the overall sound level. This allows one
to choose the frequency weighting, as will be discussed later.
The application to AIV can be seen when considering two
valves that generate the same internal sound level inside a pipe,
but the first valve produces its highest noise levels at
frequencies where the transmission loss is lowest. Then the
first valve will produce higher pipe wall vibration levels and
subsequently have a higher AIV risk.
The current IEC standard has a default spectrum for the
internal noise. The accuracy of the default spectrum for
pressure relief valves needs to be evaluated through testing.
However, the IEC standard does allow for measured spectra to
be used. And thus, if the internal noise spectrum from a
pressure relief valve is not adequately represented by the
default spectrum, then the more accurate measured spectrum
can be used so that the other advantages of the IEC standard
can be used.
60
55
50
Data
45
TL Model
40
0.01
0.1
Thickness/Diameter
2U 3 apf
Sp( f ) =
()
(5)
2fa
U
(6)
Figure 216 shows curves generated from test data and from
Norton15 for various piping elements. The noise generated by a
piping element can be calculated using Eqs. 5 and 6
S ( f )
L pi ( f i ) = 10 * log10 p 2 i
p
ref
(7)
N (L L )/10
L peT ( f i ) = 10 log10 10 pin Drn + TL( f i )
n=1
(9)
LDr = 0.06
r
Di
(8)
(FSRF SCF p )
(p )
SPLREV = 10 log10
surf
ref
(10)
(p )
(p )
L peT = 10 log10
surf
(11)
ref
EXAMPLE DATA
The primary data set available to evaluate an AIV
assessment approach is the Carucci-Mueller data1. However,
because the paper provides insufficient details regarding the
valves, piping geometry, and piping fabrication standards,
assumptions need to be made to use the data set. It was
assumed that each case had a line size standard globe valve and
the piping geometry at the AIV assessment point and the
fabrication quality was not considered. The results will be
different if the valves were of different design and not line size
valves.
Figures 4, 5 and 6 show the calculated A-weighted, Cweighted, and un-weighted overall sound pressure values
calculated at the pipe wall surface. The figures have cases
which had known failures as red circles and cases which had
not yet failed as black circles. The figures also have a red
shaded band between 125 and 130 dB, which is proposed as a
design limit. This band is set in the region of transition
between the failures and non-failure points. The band is used
rather than a sharp line, because the uncertainty in the sound
calculations is 5dB. Thus the red zone is used to indicate the
band where failure is likely and that other information, such as
past experience can be used to estimate the susceptibility of the
design to AIV failure. More work will be needed to define the
center value and range of the red band.
The difference between the A-weighted and C-weighted
values are particularly noticeable at one failure point, while
there is little difference between the C-weighted and unweighted values. The one failure point around 600 mm pipe
diameter, is below the criteria in the A-weighted plot (Fig. 4)
and in the red zone of the C-weighted plot (Fig. 5). Further
review of the flow data shows that this point has the highest
kinetic energy in the pipe flow as defined by the Energy
Institute Guidelines.4 Thus, one would expect the potential of
FIV damage in this case. It is hypothesized, that this shows that
the failure in this one case was likely a combination of AIV and
FIV. Further examples and testing are needed to determine if it
is possible to differentiate as well as perform a combined AIV
and FIV assessment by comparing the results with different
frequency weighting.
CONCLUSIONS
A strain based framework is proposed for AIV evaluation
of piping systems that is focused on using the prediction
technology contained in international standards. There are
several pieces of the framework that need to be further
developed and tested. The procedure is more complex than the
current AIV assessment approaches, but has potential to offer
more means to include the influence of various factors, and thus
reduce the safety factor required in the evaluation criteria.
The framework is based on the sound pressure on the
surface of the vibrating pipe wall being proportional to the
surface strain. This relationship allows one to bring the
evaluation approach into a fatigue based evaluation which
allow other evaluation tools such as fatigue strength reduction
factors and stress concentration factors to be utilized.
11
140
Failure
No Failure
135
130
125
120
115
110
0
200
400
600
800
1000
140
140
Failure
No Failure
135
130
125
120
115
Failure
No Failure
135
130
125
120
115
110
0
200
400
600
800
1000
110
0
200
400
600
800
1000