A Brief History of

Control Valve Noise Prediction

Herbert L. Miller, Consultant , Laguna Hills, California

The development of a physics based prediction model for the

noise generated by control valves is reviewed. The history covers
a 40 year period of experimentation and development through an
industry wide cooperation via standards committees in the United
States and Internationally. The history illustrates the multi-year
time frame associated with the development of industrial stan-
dards for complex processes. The process involved manufacturers
and users with their separate interests but with a common goal of
a good noise prediction standard. A brief description of the noise
prediction model is presented.

Significant noise can be generated by control valves, particularly

in the power industry during the startup and shutdown of these
complex boiler-turbine systems. Control valves are used to control
flow and pressure of a fluid to maintain a desired output. They are
also critical control elements for bypassing various heat exchang-
ers and the turbine during thermal transients. This requires high-
pressure letdown. Pressure reductions as high as 25 MPa (3600
psi) to a vacuum in the condenser are required. Hot water flashes
and cavitates while steam jets will whistle and screech, causing
noise levels to exceed 115 dBA.
The Occupational Safety and Health Administration drafted
noise exposure rules in the 1970s and finalized the rules in 1983.1
For example, a worker could not work more than eight hours in an
environment where noise levels exceeded 90 dBA, and a company
was required to initiate noise control administrative procedures
if the noise exceeded 85 dBA. This created a need for control
valve manufacturers to provide noise estimates for their designs
to comply with engineering specifications that limited the noise
level in a given application.
The manufacturers scrambled with many noise tests and limited
theory to develop prediction models for their products. The bulk
of the theory was based on Lighthill’s paper2 dealing with noise
created by freely expanding jets. However, the noise created by a
control valve propagates into the confined downstream piping,
which causes the development of a very complex acoustic field.
The high-pressure metal piping can have wall thicknesses that
vary from 6-100 mm (1/4 inch to 4 inches) or more. Pipe diameters
are usually less than 600 mm (24 inches) but can exceed this size
significantly for some systems.
These early prediction models were mostly empirical, using vari-
ables that influenced noise such as pressure drop across the valves,
flow rates and fluid properties. The models were very limited and
applicable to the designs that were tested. The ISA Handbook of
Control Valves3 published early versions of these empirical models
Figure 1. Comparison of different manufacturers’ noise prediction
in 1976. The handbook used input from four different contribu-
methods (1983): a) Sound pressure level of steam at constant mini-
tors, which was a reflection of the amount of new investigations
mum pressure drop at temperature conditions; b) Sound pressure
and development that had taken place and to present a “balanced
level prediction of steam at constant flow of 215 pph.
picture of this involved subject.”
With each control valve manufacturer having its own noise the pressure letdown process, there was a motivation to develop
prediction model, it was quite frustrating for plant designers and other designs. (The part called “Trim” refers to the internal part
valve users to arrive at consistent predictions of the noise levels. of the valve that causes the fluid pressure drop.) These designs
Additionally, there were more than 40 different valve manufactur- focused on reducing the fluid jet energy exiting the valve Trim by
ers capable of supplying valves to a given plant, which resulted either splitting the jets into smaller components and/or reducing
in a lot of uncertainty and varied results for what were expected the velocity of the jets by forcing the flow through more tortuous
to be similar valve designs. pressure letdown paths.
This uncertainty was illustrated by a study comparing the One of the earliest versions of valve Trim that used this concept
prediction of noise for valves with the same sizing conditions as of fluid jet energy control was patented in 19705 and is illustrated in
illustrated in Figure 1.4 The varying flow and inlet pressure were Figure 2. The flow pattern consists of multiple flow paths. Within
typical of valves handling steam in the power industry. each flow path, there are multiple right-angle turns, as flow is
As the manufacturers became more enlightened about the im- forced up and down through “punched” holes in adjacent disks.
pact of the noise created by the jets exiting the valve Trim during Each turn causes a pressure drop, and by adding enough turns, a

www.SandV.com MATERIALS REFERENCE ISSUE 13

Figure 2. Multi-path, multi-stage valve trim; multiple pressure-drop
stages consist of right-angle turns in flow path.

Figure 4. Typical response and pipe cutoff frequencies.

and one meter off the pipe surface. Results would be adjusted
using the A-weighting scale to reflect the noise an average person
would perceive.
The most complex feature of the model was the transmission
loss through the downstream pipe wall. Each pipe size has differ-
ent values and numbers of coincident frequencies impacting the
transmission. For example, a 2-inch Schedule 80 pipe, 5.5 mm
(0.218 inch) wall thickness, has six coincident frequencies; an
8-inch Schedule 40 pipe, 8.2 mm (0.322 inch) thickness, has 68 co-
incident frequencies. A typical pipe response is shown in Figure 4.
Features of the theoretical model were first published by Bau-
mann in 1982.7 The Pennsylvania University results were pub-
Figure 3. Control valve noise model and tested configuration. lished by Reethof in 1985.8 The model was used in a draft standard
for Aerodynamic Noise Prediction by ISA in 1982 and circulated
lower velocity jet exiting the trim can be achieved. for preliminary use and critique. The standard included tables
The flow in Figure 2 is directed toward the outside diameter of of pipe transmission loss in one-third octaves from 25 to 20,000
the trim. However, this could be designed to have flow from the Hz for pipe sizes from 1 to 24 inches of three different pipe wall
outside to the center of the trim if needed. To achieve the desired thicknesses for each pipe size and for pipe fluid Mach numbers
flow through the valve, an appropriate number of disk sets are of zero and 0.3. Predicting the noise required the user to make
stacked on top of each other. The flow is then controlled by a valve numerous interpolations and sum the noise for each of the octave
plug moving up or down through the center, exposing or hiding bands. Accuracy for most valves was estimated to be within ±5
flow paths. There have been many variations to this approach dBA. The valves considered in the model had a single pressure drop
throughout the years. trim like that shown in Figure 3. The standard was not applicable
Due to the uncertainty associated with various valve manufactur- to multi-path, multi-stage type trim like that shown in Figure 2.
ers’ noise predictions, the ISA Control Valve Standards Committee The ISA prediction standard caused a lot of confusion because
arranged a research project in 1974. (The Instrument Society of of its complexity and difficulty in carrying out the laborious
America is now the International Society of Automation; Research calculations. This was a period before the personal computer, so
Triangle, NC). most users had to program this procedure on a large mainframe
The project’s objective was to provide control valve users with with limited availability because of other computing demands. As
a single noise prediction method that was based on theoretical a result, the ISA committee made some simplifications by creating
fundamentals of the physical process. Multiple control valve five separate jet Mach number regimes and providing the complex
manufacturers contributed to the research that was conducted by pipe transmission loss functions in a number of representative
Pennsylvania State University.6 The research included testing of a formulations. The first approved standard was published in 1989,
simple valve design (see Figure 3) so that the methodology could be seven years after the first draft.
supported with as few empirical factors as possible. The key factors With this new noise prediction model, the users could calculate
that were established were acoustical efficiency and the amount of an expected noise level, which was frequently different from what
noise generated that exited the valve into the downstream piping. the control valve manufacturers would state. The manufacturers
Acoustical efficiency was on the order of 0.04 percent of the total needed to make adjustments to the model to improve the accuracy
jet kinetic energy, and only one-fourth of the total noise generated for their products to avoid possible warranties associated with not
was typically propagating into the downstream pipe. meeting specification requirements. The prediction differences
The resulting theoretical model for the valve tested consisted were debated, and both parties eventually moved on to other
of four parts: important matters. However, uncertainty still existed, since there
• Calculating total jet kinetic energy exiting the valve pressure was a lack of confidence in the noise prediction methodology.
reducing trim. ISA through its U.S. representative to the International Electro-
• Estimating the amount of this energy that was converted to technical Committee (IEC) urged the Control Valves Subcommittee
acoustical energy. under IEC 65B to adopt the more accurate physics-based noise
• Determining the amount of acoustical energy that was transmit- prediction method. The same confusion caused by each control
ted through the downstream pipe wall. Noise radiated from the valve manufacturer having a different noise prediction method
bulky valve body was considered negligible. also existed on an international level. There was a German Stan-
• Calculating how much of the noise transmitting through the pipe dard, VDMA 24422-1989, for noise prediction, but it relied on a
was radiated to a listener one meter downstream of the valve number of experimental factors that had to be provided by each

14  SOUND & VIBRATION/JULY 2017 www.SandV.com

manufacturer. The international committee resisted the ISA ap- jet. The vapor bubble collapses as the static pressure recovers to
proach because it did not include the multi-path and multi-stage downstream conditions. The bubble implosion can cause signifi-
low noise valve trim designs. cant noise and damage.
An IEC 65B task force was formed to create a model for these International noise prediction standards for control valves
low-noise designs using the same theoretical principles. The key now exist that are based on solid theoretical fundamentals. This
to the model expansion was the ability to predict the pressure provides a platform for continued improvements as the industry
prior to the fluid exiting the last pressure drop stage. Knowing gains additional experience.
the pressure drop across the last stage permitted predicting the jet
properties exiting the trim. The expanded model was evaluated References
using test data from five different control valve manufacturers and 1. Department of Labor, OSHA, “Occupational Noise Exposure,” Hearing
Conservation Amendment; Final Rule, Federal Register, 48, 1983. This
accepted with some limitations and larger error bands than for the
Regulation follows from the US National Environmental Policy Act of
single stage valves. 1969.
The IEC 60534-8-3 standard was published in 19959 and en- 2. Lighthill, M. J., “On Sound Generated Aerodynamically, Part 1 General
hanced to include piping expanders and orifices downstream of Theory,” Proceedings of the Royal Society of London, 211A, pp. 564-587,
1952.
the control valve in 1997 and 2000. Minor improvements in the
3. ISA Handbook of Control Valves, 2nd Edition, Editor in Chief J. W.
prediction standards were made in 2007, with the ISA and IEC Hutchison, Instrument Society of America, 1976, Research Triangle
standards being the same except for the differences resulting from Park, NC.
Imperial and Metric units. 4. Shea, A. K., “A Comparative Study of Sound Level Prediction Methods
for Control Valves.” Noise-CON 83 Proceedings, March 1983, pp 21-26.
The latest IEC standard was published in 2010 and was signifi-
5. Self, Richard E., U.S. Patent Office, High-Pressure Fluid Control Means –
cantly improved by returning to summing the noise level at each Patent 3,513,864 and High-Energy Loss Fluid Control – Patent 3,514,074,
one-third octave band as was originally drafted in 1982. Other both filed in 1968 and issued May 26, 1970.
improvements included revised acoustical efficiency functions 6. Valve Manufacturers Collaborative Program, 1974 - 1984, Pennsylvania
State University, Penn State Noise Control Laboratory, University Park,
and parameters representative of different valve types. With the
Pennsylvania.
advancement in computing capability, anyone can carry out the 7. Baumann, H. D., “How to Estimate Aerodynamic Valve Throttling Noise:
calculations on their own computer with a few inputs to arrive at a A Fresh Look,” Paper C182-902, ISA/82 Conference, Philadelphia,
good estimate of the noise level for various fluid and valve geometry October 18-21, 1982,
8. Reethof, G., Ward W. C., “A Valve Noise Prediction Method Based on
configurations. Manufacturers still have some flexibility to adjust
Fundamental Principles,” ASME Journal of Vibration, Acoustics, Stress
the calculation to best fit their products, usually in the form of and Reliability, ASME, New York, 1985.
changes to acoustical efficiency factors and more accurate predic- 9. International Electrotechnical Commission, IEC 60534-8-3, Industrial-
tions of last-stage pressure drop associated with the exiting jets. process control valves – Part 8-3: Noise considerations – Control valve
aerodynamic noise prediction method. 2010. The ISA harmonized noise
The principles used for the aerodynamic noise prediction have
prediction standard, ISA 75.17 will be released in the future.
now been applied to liquid flow with the publication of the IEC 10. International Electrotechnical Commission, IEC 60534-8-4, Industrial-
60534-8-4 standard for hydrodynamic noise prediction in 2015.10 process control valves – Part 8-4: Prediction of noise generated by
Liquid flow calculations consider the noise due to turbulence and hydrodynamic flow, 2015.
cavitation. Cavitation is a process where the fluid’s vapor phase is
created within the jet due to low static pressure in the high-speed The author can be reached at: herb.miller@imi-critical.com.

www.SandV.com MATERIALS REFERENCE ISSUE 15