You are on page 1of 4

The IEC Aerodynamic Valve Noise

Prediction Standard Revisited


Hans D. Baumann1 and Ed Singleton2
1
Consultant, H.B. Services Partners LLC
2
Consultant, Koso Kent Introl

Introduction of the existing documents and for a number of reasons the ISA
standard was chosen. One of the main reasons for this decision was
Engineering standards whether of national or international status the committee’s aim to have an international standard that anyone
are, by the rules of their responsible institutions, subject to review could use for the prediction of noise levels without recourse to test-
or “maintenance” periodically, usually every five years. It was to ing or requesting test data from the manufacturer. More precisely,
comply with this rule that the IEC control valve noise prediction the aim was to present a method of noise prediction that requires no
standard 534-8-3 came up for review in the year 2000 and again in more information than the valve sizing coefficients and the service
2005. The review of 2000 left the standard relatively unchanged, the conditions. The ISA standard offered this facility whereas the VDMA
team of review experts having decided that it was fundamentally method at that time required the value of the efficiency factor η,
satisfactory. and other factors, to be determined by testing. This would involve
The review of 2005 has extended into 2007/8 with a working the specific valve manufacturer supplying these values or the user
group of noise experts still scrutinizing the standard, even as this paying a specialist laboratory to carry out the tests.
paper is being written. That the standard is currently being revised is In due course the working group produced the IEC draft standard,
common knowledge throughout the world-wide national standards based on the ISA standard, and this was presented to the 1992 meet-
committees so there is a certain amount of speculation, outside the ing of IEC subcommittee 65B/WG9 in New Orleans. At that meeting
ranks of the IEC, about the proposals that will eventually emerge an acoustic expert, Erik Tromp from Shell (The Hague), compared
from the IEC working group after approval has been given by the full predictions using the IEC proposed draft with test results for more
IEC subcommittee 65B/WG9. This could result in extensive changes than 100 examples, demonstrating that 80% of the predictions fell
or no changes or anything between these two extremes. The next within 3dB of the tests and 90% fell within 5dB. At that meeting
stage will be the circulation of the draft to the various national the draft was approved and the IEC aerodynamic noise prediction
standards committees who will be asked to forward comments for standard 534-8-3 was published in 1995.
further consideration by the working group 9 of the IEC subcom-
mittee 65B.
Whilst the proposed revisions are being discussed at international
and national levels, users should maintain a close liaison with their
The present day status of IEC
national committees if they would like to have an opportunity of
commenting. It is the normal procedure for national committees
534-8-3
to forward selected comments to the appropriate IEC committee. The former ISA standard SP 75.17 on the Prediction of Aero-Dynamic
Because this standard, along with the hydrodynamic noise predic- Valve Noise has now been replaced by the international IEC stand-
tion standard, is the most complicated in the IEC 60534 control valve ard 60534-8-3, and this, like its predecessor, has served the process
series, users should be mindful of the fact that changes to computer control industry well. Its instant appeal and widespread popularity
programmes, technical literature and training programmes could derive from its sole dependence on scientifically based calculations
be difficult and expensive. Radical changes to the standard should avoiding any recourse to testing, thus enabling it to be used by any-
therefore be fully justified. one for valves of almost any manufacture. It is quite accurate con-
All users of this standard should have an interest in its future, but sidering the apparent complexity of the underlying fluid-mechanic,
a look back at its origins and the rationale that led to the standard in thermodynamic, and acoustic laws involved. Even testing conducted
its current form could be useful in focussing direction when ponder- in refineries by the Shell Oil Company, and considering the ambient
ing over proposed changes that could have a significant effect. influences which it involved, showed that typically eighty percent
of all tests conducted with a variety of valves, including low-noise
types, lay within plus or minus three dB of the predicted range, while
Historical note over ninety percent fell within plus or minus five dB. Similar results
were shown in published test data by such well known companies
It was not until 1990 that the IEC decided to turn its attention to the as Fisher Controls, Introl and Masoneilan1, to name a few.
preparation of a valve noise prediction standard. Two quite different The standard is specified by most of the major operators in the oil
methods were already available: one from the German VDMA – ref and gas industries. It has been put to the test in the design stages
24422 and another from the USA ISA-std S75.17 (approved as an of plants where compliance with strict local environmental condi-
ANSI std in 1991). Rather than create a completely new standard, tions limiting the level of noise emissions was a legal requirement.
subcommittee 65B/WG9 decided to model its standard on one Practical experience with IEC 534-8-3 in these demanding cases

www.instmc.org.uk Measurement + Control Vol 41/5 June 2008 • 143


has proved its reliability. The IEC method is also recognized by two Lpi = 10 log (3.2 x 109 x Wa x ρ2 x c2 / Di2) (dB).
world-renowned Noise Control Handbooks2, 3. Here ρ2 = density of the gas down stream (kg/m3), c2 = sonic velocity of
However, such accuracies can only be obtained if important valve the given gas (m/s), air is 334 m/s. Di = inside diameter of pipe (m).
parameters such as Cv, Fl, and Fd are known and correctly used.
These parameters do vary with valve travel and this has to be taken 5. Since we want to know the sound pressure outside of the pipe,
into account. It is, for example, not correct to use an Fd coefficient we now have to subtract the transmission loss TL of the pipe wall,
measured on an eight inch ball valve and then to use the same value where (somewhat simplified to fit most cases):
for a two inch valve. After all, one would not expect to calculate a
flowing quantity of a valve while using the wrong Cv number and TL = 10 log [ (7.6 x 10-7) x ( c2 / fp x tp)2 / (ρ2 x c2 / 415) + 1] dB
then blame the ISA sizing standard for the resultant error.
Yet, there are areas within the current noise prediction standard, which where: tp = the wall thickness of the pipe (m) and fp the frequency
could stand improvements. Whilst these discrepancies, which have been at which the internal sound pressure Lpi is at it’s maximum (Hz). Now
noted over the years, are relatively minor and affect only a small number this is the trickiest part. Miscalculation of fp typically creates most
of valves they nevertheless should receive some attention. of the prediction errors, therefore, Fd has to be known in order to
calculate the correct jet diameter Dj.
Suggested improvements are:
A. A correction of the acoustical efficiency factor in the sub-sonic fp = 0.2 x Uvc / Dj (Hz).
velocity area, favouring enhanced dipole noise sources.
B. Adjust the changes in the pipe transmission loss for cases where 6. Now all we have to do is calculate the Sound Pressure Lever
there is a large pipe, and a high peak frequency producing trim. outside the pipe, and typically at one metre from the pipe wall
C. Correct the A- weighting factor for cases where fo the co-incidence (and for low outlet pipe velocities):
frequency (where most of the noise escapes) lies above 10,000
Herz (the audible limit). This typically happens if the gaseous fluid LpAe,1m = ΔA + Lpi + TL – 10 log [ Do +2 / Do] in dBA
has a high sonic velocity (Helium for example).
where A is a correction for the ΔA – weighted frequency correction,
assumed to be 5 dB. Do = outside diameter of the downstream pipe
How does the current IEC method (m). Note: Don’t be confused by “ + TL, ” TL really has a negative value.

work? Suggested improvements:


From various sources – see refs. (these if adopted will give minimum
Reading the complete IEC prediction standard can be confusing due disruption to the current standard.)
to the many equations. This in turn has created misunderstandings
so the reader is now given a “scaled-down” version in six easy steps A. To improve the slope of the acoustical efficiency factor in order
to show the underlying simplicity (at least for the vast majority of to get better results for low pressure drops (low Mach numbers
all valves). ). This can be done by changing the slope of η from MVC3.6 to
MVC3.0 in noise regime 1 (sub-sonic) by giving more weight to the
1. First calculate the velocity of the gas jet coming through the dipole noise sources. This would increase the predicted noise at
orifice, where; Uvc= 33 m/s (ΔP/P1≈0.005) by 6 dB and for Uvc=140 m/s (ΔP/
P1≈0.05) by 2.9dB for example. This revision would eliminate
Uvc = 4 x m / ρ x Dj2 x Π (m/sec); where: m = mass flow most of the low Mach number errors exhibited in reference 1.
(kg/s), ρ = density (kg/m3), Dj = diameter of jet in meters given as Incidentally, this modification was originally recommended in ref 4.
0.0046 x Fd x ( Cv x FL) 0.5.
B Improve the transmission loss calculations for typically low
2. Calculate the mechanical power that is converted in the valve noise valves installed in large pipes (8 inches and larger). This
due to pressure drops: improvement will account for the noise increase when the
peak noise frequency inside the pipe is more than 2 octaves
Wm = m x Uvc2 / 2 (watts). (4 times) above the pipe’s internal coincidence frequency fo.

3. An acoustical efficiency factor is used to calculate the sound To explain: as one can see from Figure 1, the internal sound pres-
power created by jet turbulence inside the downstream pipe: sure does not decay at a constant rate, but changes slope when
the pipe internal frequency reduces below about 0.25 fp, from
Wa = η x Wm x rw x FL2 (watts); 20 dB per octave to only 10 dB. This change is not recognized in
the current standard but could be included by introducing an
where: η is the acoustical efficiency factor which has a value of 0.0001 additional factor called ΔTLfc where;
at sonic velocity (mach 1) and has a max value of 0.001 at Mach 2.
Below Mach 1 it decays at a rate of MVC3.6 (for practical reasons this ΔTLfc = 10 log ( fp / fr) – (10 log (fo gas / fo air))*.
may be modified to MVC3.0). Factor rw for globe valves has a value of ΔTLfc will then be added to LAep,1m.
0.25. FL is the valve’s pressure recovery factor. NOTE: ΔTLfc has to be a positive number..

4. The next step is to convert the sound power into the internal * This part is applicable only to monatomic gases.
sound pressure Lpi:

144 • Measurement + Control Vol 41/5 June 2008 www.instmc.org.uk


In the test data shown in Figure 2, the medium was Helium having
a speed of sound of 972 m/s. the two inch pipe used therefore had
a coincidence frequency fo of 22,500 Hz.
Here: ΔA = 5 – 10 log ( 22,500 / 6000) = 0. In this case, there is no
A-weighted correction as Kurt Roth proposed in his original paper*.
This will lower the predicted value by 5 dB, thus making the error
between prediction and test practically zero as the graph indicates
Figure 3 shows a typical external sound pressure profile with the
A-weighted additions per given frequency superimposed to aid in
understanding the problem6.
Argon, another test medium, has a speed of sound of 300m/s
(957f/s). Here ΔA=5-10 log (6944/6000=4.4dB), which agrees with
test data.
Figure 1. Test data showing the internal frequency dependent sound
decay.

For example, if fp = 18,000 Hz (cps) and fr = 5300 Hz (12 inch pipe,


medium air), then ΔTLfc = 10 log (18,000 / 5300) = 5.3 dB. This will
increase the calculated external sound level by 5.3 dB. The above
example is shown graphically in Figure 1.

Note: ΔTLfc cannot be less then zero..


Such a modification has been requested by the Italian National
Committee for WG 9.

C. Improve the A-weighted correction. The A-weighted external


frequency correction of 5 dB is a good average for the wide
majority of all cases since it saves additional tedious calculations.
Such calculations are also subject to the rather erratic behaviour Figure 3. Test data on a 3” rotary valve with low noise trim, fp = 20 kHz,
of the pipes transmission loss around fo. However, Kurt Roth5, LpAe,1m was calculated as 99 dBA. This agrees with the 94 dB measured at
formerly of Masoneilan, found that gases having a high speed 7500 Hz plus 5 dB for ΔA.
of sound, such as Helium, created coincidence frequencies
in small pipes, that were well above the 10,000 Hz audible
frequency level. This problem could be alleviated by making the
A weighting correction variable in cases where fp is higher than
Conclusions
fo, such that; Realistically, over a period of 12 years, the IEC aerodynamic
noise prediction standard has been used successfully by valve
ΔA = 5 – 10 log (fo / 6,000). Note: ΔA cannot be less than zero. manufacturers, plant contractors and operators. The step-by-step
calculations follow a logical pattern enabling those interested to
fo = fr x Co gas / 4 x Co air in Hz. trace the derivation back to the original work by Lighthill, Powell
and others in the 1950s. It is to be hoped that this transparency is
not lost in any future revisions. This could be lost if some ideas being
promoted from within and without the official working group were
to be accepted. To quote one such example: the combination of
the acoustic power ratio rw with the acoustic efficiency factor η has
been suggested. It is possible to do this without changing results
but it would obscure the origin of η and with rw having disappeared
the user would be ignorant of the amount of sound power that
the standard was apportioning to radiation into the downstream
pipe. Another equally disturbing suggestion is the idea of using the
Strouhal Number to define the jet diameter. This is not a good idea.
The Strouhal Number is the dimensionless ratio of the product of
frequency and diameter to jet velocity and should not be used to
define jet diameter.
It serves no purpose to dismantle the current standard if only for the
sake of changing nomenclature and improving accuracy for infrequent
and unusual service conditions where the required improvement can
be achieved by small changes as explained in previous paragraphs,
leaving the basic form of the current standard unchanged.
Figure 2. Two-stage drilled hole cage test fluid - Helium This standard features prominently in the computer sizing and
specification software of most of the organisations concerned with

www.instmc.org.uk Measurement + Control Vol 41/5 June 2008 • 145


control valve technology. For these organisations to be compelled to their teams of experts that it is not a textbook they are writing, but a
make radical changes to software and recirculate them throughout standard that will be used by more non-experts than experts.
their networks would be time consuming and expensive. An ad The IEC subcommittee 65B/WG9 may well concur with satisfied
hoc review does not indicate that the standard in its present form users of the standard and reissue it with nothing more than minor
is giving serious problems. One control valve manufacturer, with changes, but if it should decide that radical changes are necessary
over four years experience in the use of this standard in its work it would be reassuring to all users if the revised standard were to
on offshore platforms and land-based production and processing be verified through testing by an independent concern in a similar
plants, testifies to its extensive capabilities and universal acceptance. manner to the acceptance tests imposed on the current standard
There has been no evidence from a broad field of users claiming that prior to its publication in 1995.
the method is not “fit for purpose”.
It should be of concern to all users of the standard that its most References:
appealing feature of “prediction through calculation” is retained, 1. Henry W. Boger, Improving prediction of control valve noise, InTech, August 1988, pp.61-62.
along with its clear technical procedures. The idea of vendors being 2. I. L. Vėr and L. L. Beranek, Noise and Vibration Control Engineering, 2nd Edit. John Wiley and Sons, 2006.
allowed to determine acoustic efficiency factors by testing as an 3. Malcolm J. Crocker, Handbook of Noise and Vibration Control, John Wiley and Sons, 2007.
alternative to the calculation method quoted in the standard has 4. E. W. Singleton, Understanding IEC 534-8-3 in Control Valve Aerodynamic Noise Prediction, Measurement +
been promoted in some circles but this must be avoided except in Control (UK) , March 1999, Vol. 32, Issue 2,
cases of exceptional valve designs which should not be within the 5. Kurt W. Roth, Validation of the Control Valve Aerodynamic Noise Prediction Method for Light and Heavy Gases,
province of this standard. The introduction of acoustic efficiency Proceedings of the Third Biennial World Conference on Integrated Design & Process Technology, Vol. 6, July, 9,
factors determined by valve manufacturers and others would 1998.
introduce confusion and uncertainties when comparing predictions 6. Hans D. Baumann, and Heinfried Hoffmann, Method for the estimation of frequency-dependant sound
from different sources. The practical experience of many users of the pressure at the pipe exterior of throttling valves, Noise Control Engineering Journal, Vol. 47 (2). March-April,
standard dispels any ideas that it is in need of serious maintenance. 1999, pp. 49-55.
Some minor imperfections have been noted but, as explained in a
previous section, these can be rectified without demolishing the Baumann and Singleton were members of the IEC subcommittee
effectual and well-proven methodology of the current standard. 65B/WG9 when the first issue of the noise prediction standard was
Although valve noise prediction is a complicated subject the prepared and published. They were also members when it was first
current standard presents it in a simple form with a reasonable revised in the year 2000.
compromise between academic accuracy and ease of usage.
Chairmen of standards committees have been known to remind

146 • Measurement + Control Vol 41/5 June 2008 www.instmc.org.uk

You might also like