PIPING VIBRATION: CAUSES, LIMITS & REMEDIES

PART 1: NATURAL FREQUENCY OF A PIPE

Why do pipes gallop?

Maybe galloping isn’t the right word; I would prefer to say the pipe is vibrating. But why is it vibrating
so much? One word – resonance.

If you seat a child on a swing and you give a gentle nudge to the swing every time the child swings back
towards you, you will quickly be able to get the swing higher and higher. You only need a very small
force because you waited until the swing had reached its limit of movement (the limit of the swing) and
only then did you give that gentle push. But you did it EVERY time the swing moved back towards
you. A couple of interesting things are happening here.

(1) You are pushing at exactly the same frequency as the frequency at which the child is swinging. Don’t
forget that if you are troubleshooting a pipe vibration problem the natural frequency will depend not
only on the steel of the pipe but also the mass of the fluid inside.

(2) You are pushing at the same position in the swing cycle – in other words you have “locked phase”
with the swing.

With these two “interesting things” it means you are in RESONANCE with the swing. When a pipe is
vibrating heavily it is almost always because there is a resonance issue. The swing has a natural
frequency because it has a period of oscillation that depends on the mass of the swing and the length of
the pendulum. This is the reason why a pendulum is used on a clock. If you can work out the period of
the cycle back and forth, then you know the frequency.

If the time period of oscillation is one second then the frequency is one per second or 1Hz. If the time
period is half a second then the frequency is 2Hz and so on. By the same reasoning the pipe has a
natural period of oscillation and so it has a natural frequency. The natural frequency of the pipe
depends on its stiffness and its mass; the stiffer the pipe the higher the frequency, the more mass the
pipe (including contents) has, the lower the natural frequency.

To calculate the natural frequency of a pipe with rigid supports use the following formula:

Where:

f = natural frequency of the pipe (Hz)

E = Young’s modulus of elasticity (200GPa or 30E6psi for steel – approximately but close enough)
I = 4th polar moment of inertia for the pipe (0.049*[OD^4-ID^4]) in inches or metres

µ = mass per unit length of the pipe (remember to include the mass of the fluid) lbs/inch or kg/m

L = distance between pipe supports (inches or metres)

Let’s find the natural frequency of a 12” (300mm) pipe made of A-106 GrB schedule 80 that is first
empty and then filled with water. I’ll use SI units to make the math easier. Pipe supports are 5m apart.

Given:

Pipe OD = 323.8 mm, Pipe ID = 288.84 mm,

Mass/length empty= 132.05 kg/m,

Mass/length full = 132.05 + π(0.28884)^2/4 * 1000 = 197.57 kg/m E = 200 x 10^9 Pa

(reprinted from TIOGA pipe chart)

Calculate I

Natural Frequency:
Now you know the natural frequency of the pipe you ask your vibration techs to take a vibration
measurement on the pipe when it’s in operation. If the frequency is the same as your calculated
frequency, then the pipe has a resonance problem and the next step is to identify what is the force that is
exciting the natural frequency.

Check out the next STEP - STEP 2 - Calculate VIV Vortex Induced Vibration affecting the pipe.
PIPING VIBRATION: CAUSES, LIMITS & REMEDIES

PART 2: CALCULATE VIV VORTEX INDUCED VIBRATION

In part 1 we determined the natural frequency of the pipe so now we know the pipe natural frequency
but you are not standing there pushing it – so what is? It could be any of a number of things:

(1) Vibration at the same frequency coming from a pump or compressor (usually speed related). This
could be caused by unbalance, misalignment, something may have come loose or just about any fault on
a machine that causes a vibration at the same frequency as the natural frequency.

(2) Flow induced vibration. Now this could be from the internal flow of the fluid through the pipe or
even from wind flowing across the outside of the pipe. We have all seen the effect of wind induced
vibration on street lamps or poles so when the wind hits a particular speed the pole starts to
sway. That’s because the oscillation force associated with the wind is a function of the pipe outside
diameter and the wind speed so the wind vortexes or swirls on the downwind side of the pipe and the
vortexing induces KARMAN vibration.

Check out the link for more information

I usually start with the easiest option. Have a look around the pipe and see if there is any rotating
equipment that has a run speed (or a harmonic of run speed) that is very close to the pipe natural
frequency. If there is we can either change the speed of the machine (if possible, because that is the
easiest option) or change the natural frequency of the pipe. An easy way to change the natural frequency
of the pipe is change the value of L – in other words change the location of the pipe supports or maybe
just add another support. If you add another support be careful that you put it at a location of high
amplitude – in other words an antinode of vibration.
You can see from this image that the modifier we used (a=22.4 ) to determine the natural frequency of
the pipe in Part 1 is only applicable to the first mode of vibration of a “clamped-clamped” beam or
pipe. If you install another pipe support halfway between existing supports but the pipe is vibrating at
the second mode (a=61.7) the amplitude of vibration will be unaffected. I usually hammer in a stout
piece of wood as a (very) temporary measure to see if that indeed reduces the vibration.

If you find that there is no rotating equipment nearby that could affect your pipe we could see if there
are any other possible causes and an easy one to check is vortex induced vibration.

For the same reason that a flag flutters, a pipe (or any object) will experience an oscillatory force when
placed in a fluid flow. As the wind flows across the pipe there are tiny differences in air pressure from
one external side of the pipe to the other so the wind finds slightly less resistance on one side and more
wind flows towards the lower pressure. As more wind flows towards the lower pressure side that side
experiences an increase in air pressure so the flow of wind flips over to the other side. The other side
experiences the increase in air pressure so the flow flops back again. The flow of wind is now flip
flopping back and forth causing a transverse oscillating force on the pipe.
By Cesareo de La Rosa Siqueira - Copyrighted free use

Going back to our flag analogy, the pipe flutters as the wind passes the flagpole and vortexes on the
downwind side. The vortex is travelling along the flag and the flag “flutters”

We, though, are interested in what is happening to our pipe. According to Strouhal and Karman there is a
distinct relationship between the speed of the wind, the diameter of the pipe and the frequency of the
oscillating force.

St = fD/V

Where

f is the frequency (Hz)

D is the diameter of the pipe

V is the wind velocity.

St is the Strouhal number. This does vary somewhat with Reynolds number but we can assume it to be
0.22. So if we have a wind velocity of 10 m/s and we know that our pipe has an OD of 0.3238 m (see
Part 1 where we calculated natural frequency) the VIV frequency is 0.22*10/0.3238 = 6.79Hz. Easy isn’t
it?

The vortexing frequency effect is not limited to external wind however. You will get the same effect from
fluid flowing inside the pipe as it flows across an obstruction. So if we have a gate valve with a non-rising
stem with a stem diameter of 3.5cm and a fluid flow of 10m/s we would have a vortexing frequency of
62.8Hz. Remember that our pipe has a natural frequency of 63.77Hz which is close enough to ensure
resonance.

In our next article we will examine the effect of changing the fluid velocity and see how that affects
resonance in our pipe.
PIPING VIBRATION: CAUSES, LIMITS & REMEDIES

PART 3: VIBRATION FROM FLOW VELOCITY

In Part 1 we figured out the natural frequency of a pipe and in Part 2 we looked to see if the resonance
excitation was from a nearby rotating equipment or perhaps vortex induced vibration. If neither of these
options came close to identifying the forcing frequency we need to look at slightly more exotic causes.

Most people who work in process, power, oil & gas or refining will have come across a problem in which
a perfectly normal section of piping with no significant vibration “suddenly” starts to vibrate for no
apparent cause apart from a slight change in flow rate. But that doesn’t seem to make any sense. We
have made no change to the mass or stiffness of the pipe so the natural frequency hasn’t changed. We
have not changed any run speed of nearby equipment and even if we check VIV vibration it doesn’t even
come close to the problem frequency. What the heck could it be?

Let’s take a trip down to the train station. The express through train is coming down the track and as it
passes us we hear a definite change in pitch. A high pitch as the train travels towards us and a lower
pitch as the train moves further away. We are talking about the DOPPLER effect.

To understand Doppler we need a good understanding of noise and sound. When we speak the sound
we make travels through the air at about 330m/s. That doesn’t mean we are expelling air from our
mouth at that speed – that would be rather unpleasant. As our voicebox vibrates it creates an area of
high air density as it pushes onto the air molecules. That high density rams into the air next to it and
bounces back transmitting the energy to the adjacent air molecules. The rate at which the energy is
transmitted to adjacent air molecules is the speed of the sound. Remember from high school physics
class that the speed of sound is function of frequency and wavelength?

Where:

C is the speed of sound in the fluid

f is the frequency of the sound

is the wavelength between the high pressure pulsations of the sound

As a stationary observer of a stationary object that is making a noise the speed of sound is set by the air
density and the air pressure. The pitch or frequency is determined by the distance between the high
pressure pulsations. So let’s see what happens if we start moving the object making the noise.
When our train is stationary the sound moves away from the train at 330m/s in all directions. The
wavelength is the same as the sound travels in all directions. But when the train starts to move we have
the sound AND the train travelling in the same direction to the front of the train but moving in
OPPOSITE directions when viewed from behind the train. The speed of the sound hasn’t changed but
because the train and the sound are travelling in the same direction at the front of the train the
wavelength is compressed. Using C=fλ as the wavelength is compressed and the speed of sound stays the
same then frequency must increase. The opposite effect happens as the train moves away from us so we
hear a lower frequency or deeper tone.

The factor C is a physical value depending on the properties of the fluid. The wavelength though is
affected by the speed of the fluid flow through the pipe. The Doppler effect.

https://en.wikipedia.org/wiki/Doppler_effect

But we have a pipe with fluid travelling along the inside of the pipe. What does this have to do with
moving trains? Only that in both cases we have to think about Doppler.
Let’s move to wind musical instruments. When air is blown into a trumpet or trombone you only get a
distinct tone if you blow into the mouthpiece at a particular rate – that is why trumpeters “purse” their
lips to get the right air speed. When you get the correct air speed the wavelength becomes the same as
the length of tubing so you get a standing wave and the instrument sounds the desired note. The length
of the tubing can be adjusted on a trumpet using valves or on a trombone using the slide. In effect you
are changing the wavelength of the air and that is changing the frequency. All of us have even tried
blowing across a part empty bottle and we get a tone if the speed of the blow is “just right”.

The nearest our pipe is to the analogy of a trumpet is a section of pipe between two bends. The half
wavelength is actually longer than the distance between the bends (add about 30%). The part open
bottle analogy equates to a dead leg at a tee and in that case we have a quarter wavelength.

So now let’s combine the standing wave and the Doppler effect in our pipe. Instead of a train moving we
now have liquid or gas moving along the pipe. In effect we are causing the wavelength to change by
changing the flow velocity relative to the speed of sound. This can get us into trouble in one of two ways:
1. If you have a forcing vibration at the same frequency you get resonance of the fluid inside the
pipe. This happened to me once on a pump running at 2970rpm discharging into a line that had a length
between the discharge flange and the next tee that equated to the standing wave that had a frequency of
50Hz. Vibration at about 20mm/s rms and a bearing life of 3 months. We changed the configuration of
the discharge piping and vibration came down to less than 2mm/s and no more bearing failures.

2. If the frequency of the standing wave is close to the natural frequency of the mechanical section of
pipe you have resonance. This happened on the discharge of a large blower in China. The distance
between the discharge flange of the blower and the NRV gave a standing wave with a frequency that was
very close to the natural frequency of the piping. We moved the NRV and the vibration problem
disappeared.

Remember that standing waves can occur not only at fundamental frequency but also at
“overtones”. Don’t ignore the overtones.

We still use C=fλ to calculate the fundamental frequency so with a speed of sound of 330m/s and an end-
corrected wavelength of 10m we would have a fundamental tone of 33Hz. But as soon as fluid flows
through the pipe we have to modify that wavelength. So if we flow at 33m/s that means a 10% change in
apparent speed of sound so we would get a new frequency of (330+33)/10 = 36.3Hz.

Our pipe is carrying 22MW natural gas at 80֯C with a speed of sound of 365m/s. Our section of piping
has a length of 9.3m between bends which equates to an open/open pipe of 9.3m length which gives a
wavelength of 18.6m. Add 30% to that length for “end effect” correction and we get a fundamental tone
or frequency of 365/(10*1.3) = 15.12Hz. But we now flow gas through the pipe at 20m/s which, when
considering the Doppler effect, changes that frequency to 15.94Hz. But what makes it really worrying is
that the first overtone is 31.89Hz, the second overtone is 39.87Hz but the 3rd overtone is 63.7Hz which
is right on our mechanical natural frequency of 63.77Hz (we have ignored the mass of the gas as it is so
low) .
On very large diameter piping there is a possibility of shell wall resonance but that is quite rare and
tends to happen on trunking rather than pipework (large diameter and thin wall) so we won’t get into it
here.

PIPING VIBRATION: CAUSES, LIMITS & REMEDIES

PART 4: LIMITS FOR PIPING VIBRATION

OK – let’s recap. We have a pipe with a natural frequency and we have a force with the same frequency
and that means high amplitude vibration. So what?

If we leave a vibrating pipe in place long enough and the vibration is severe enough the pipe will develop
a crack and we get a leak. We are talking about fatigue failure. To make things easy for us there are
several versions of fatigue limits we can apply to piping and the one I will mention is API STD 618. Now
before you start jumping up and down complaining that is a standard for reciprocating compressors let
me say that yes, you are right. But this section of the standard works for all steel piping because it is
VERY conservative.

Let’s look at some of the detail.

Section 7.9.4.2.5.2.4 Piping Design Vibration Criteria

The predicted piping vibration magnitude shall be limited to the following:

a. Constant allowable vibration amplitude of 0.5 mm peak-to-peak (20 mils peak-to-peak) for
frequencies below 10 Hz (the frequency of 10 Hz is also according to ISO 10816);

b. Constant allowable vibration velocity of approximately 32 mm/s peak-to-peak (1.25 in./s peak-to-
peak) for frequencies between 10 and 200 Hz.

We need to be aware that 32mm/s pk-pk is the same as 16mm/s peak and 11.3mm/s rms. To convert
displacement to velocity:

V = 2π. Displacement. Frequency

So, 0.5 mm at 10Hz gives a velocity of 2π*0.5*10 = 31.42mm/s

Vibration measurements are almost always displayed in terms of peak-peak for displacement (total
movement) but velocity readings are only ever shown as zero to peak or even rms so don’t be confused
by the API 618 limits. To see the relationship between pk-pk, peak and rms look at the image below.
The rms. or root mean square velocity is a quantitative measure of the effective velocity and reflects the
power or energy being used to vibrate the machine mass. Peak value is the maximum amplitude seen
during the measurement referenced to zero velocity and peak to peak is a measure of the total
movement so is usually only used for displacement.
This image illustrates the movement of a pendulum. The figure above shows that at position B and C, the
velocity is zero, and at position A the velocity is maximum, first to the right, then to the left. The negative
peak velocity differs only in direction, not magnitude. The rate of change of displacement is the velocity,
therefore if D is expressed in terms of mm, instead of the usual micron, then the product 2pfD will be the
velocity in mm per second which are the units used for velocity in vibration work.

This relationship between velocity and displacement is an important factor when considering severity of
piping vibration – if you want to be accurate. I really don’t care about vibration amplitudes or movement
in pipes, what I care about is the stresses that have been imparted to the pipe. If we consider a resonant
pipe then the actual forces are quite low but the physical movement (or displacement) could be quite
high. If we consider the stress-strain diagram for carbon steel

A106GrB has a tensile stress specified at 415MPa and yield stress of 240MPa (see below).

We need to make sure the stress value due to the bending effect of the piping vibration is well below the
yield value and we can use our high school physics to do this.

Stress from bending = 8EDΔ/L2

Where:

E = Youngs modulus

D = pipe outside diameter

Δ = peak to peak displacement

L = length of pipe between supports

So let’s say our 5m pipe (between supports) has a maximum vibration amplitude of 0.5mm pk-pk. The
bending stress is

S = 8 * 200E9Pa * 0.3238m * 0.5 / 25 = 10.36MPa

That sounds quite reasonable but what happens if the vibration is 5mm pk-pk? The stress is now
103MPa. Compare that to our yield stress of 240MPa and you would think our pipe should be fine but
API 618 warns us that if there is a cyclic stress level in the piping that stress level must not exceed the
endurance limits of the piping materials. So what is the endurance level? It is the value of the stress
below which a material can presumably endure an infinite number of stress cycles, that is, the stress at
which the S-N diagram becomes and appears to remain horizontal. The existence of a fatigue limit is
typical for carbon and low alloy steels.

Looking at the image below we have a plot of applied stress against the number of cycles the metal
endured before rupturing of a carbon steel of a known UTS. It is possible to predict (very approximately)
the life to failure of the equipment if we have this plot. Our A-106 pipe steel follows the dashed line fairly
closely so we can see that if we have a stress of 103MPa the steel would be expected to fail due to cyclic
loading at about 200,000 cycles.

If our pipe is vibrating at 67Hz that means 67 cycles per second so we have a life before failure of
200,000/67 = 2958 seconds – less than an hour.
To operate our pipe safely for an extended period then the vibration MUST induce a stress that is less
than the endurance limit. And that is why the API recommendation is so conservative. But if you really
need to operate the equipment you can carry out a fatigue analysis as we have done or you could use API
579-1/ASME FFS-1 which can guide you through a step by step procedure to determine if you can use
the pipe for an extended time.

So for a final recap. If your pipe is vibrating you should do the following:

1. Find out if the vibration is resonant – that means find the natural frequency

2. If the vibration is resonant find out what is the forcing frequency – that could be a nearby piece of
equipment, external effects such as wind or fluid flow induced vibration

3. Determine if the vibration levels are acceptable – you could use API 618 for that or carry out a full
fatigue analysis.

To remedy the problem you simply have to separate the natural frequency from the forcing frequency.

To change the natural frequency you must change either the stiffness or the mass. It is usually much
easier to change the effective mass by changing the length between supports either by moving one of the
existing supports or by adding a new one.

Otherwise you have to attack the forcing frequency. If it’s a rotating machine you can change the speed,
isolate the machine from the pipework or fix the vibration issue such as unbalance or misalignment. If
the problem is flow induced or VIV you will have to do the analysis to positively identify the culprit and
make changes in flow rate, piping arrangement or even production procedures.
A Rule of Thumb for Vibration
(Velocity) Limits for Plant Piping
Systems
Posted by AnnMarie Fauske on 01.06.17

By: Jens Conzen, Director Plant Services, Fauske & Associates, LLC
Typical power and process plant systems will always vibrate under operation. This is due to various
reasons such as turbulence, acoustics, or due to the excitation from rotating equipment, for example.
Hence, it is important to monitor vibration levels to assure that the levels are not a safety concern. One
major question that arises is: At what level pipe vibrations become severe and what parameter(s)
should be measured?
It is the objective of this article to provide a simple rule of thumb to plant engineers, vibration engineers,
and managers that are facing a vibration issue at their power or process plant. Vibration issues on piping
systems can occur during commissioning of a new plant, after power uprate, or after major equipment
replacement.
Vibration data is typically captured by acceleration sensors (i.e. accelerometers – Figure 1). In that case,
it would be simplest to present the data in the format of acceleration versus time or acceleration versus
frequency. In some cases other parameters might be more important such as clearance. For example, in
rotary machinery it is not uncommon that the shaft displacement is monitored with proximity sensors
near the bearing points. Displacement becomes smaller at high frequencies and acceleration becomes
smaller at low frequencies. Vibration velocity is the derivate of acceleration and the integral value of
displacement. Hence, it lies midway between displacement and acceleration and, thus, covers a broader
central frequency range. Piping vibrations typically occur in a more central frequency range i.e. 10 to
1000 Hz, which makes vibration velocity an attractive parameter. Most importantly, it also relates to
dynamic stress. For any linear structure vibrating at resonance the following correlation can be used:

(Ref. 1)
The maximum dynamic stress equals to the ratio of maximum vibration velocity (Vmax) to the speed of
sound in the material times the Elastic Modulus (E) times some constant (the proportionality factor).
The constant is not expected to vary greatly, even over a wide range of system size, geometry, vibration
mode and frequency. NUREG-1061 suggests a range of 1 to 3 for simple practical structural elements.
The correlation works for any flexural vibration of beams (i.e. pipes) and plates with any practical
section shapes and boundary conditions so that vibration velocity becomes an appropriate parameter to
address the severity of structural vibration. What is an acceptable level? The allowable limits are
somewhat dependent on system size, layout and mode of vibration. Hence, an acceptance criterion is
best derived from a pipe stress report or seismic analysis. For cases where this information is not
readily available, a rule of thumb would be convenient to have. The best way to develop a rule of thumb
for vibration velocity as general severity criterion for pipe vibration is to correlate actual field
experience data. The cases should cover a wide range of power plant and process plant piping systems.
Before we consider field experience, let’s first take a look at the governing design standard for vibration
velocity limits for piping in nuclear plants i.e. ASME ANSI-OM3. This standard provides an allowable
zero-to-peak velocity, which is derived from the linear stress-velocity relationship above. Engineers
from EdF (Électricité de France) performed a study to verify if a general screening criterion of 12 mm/s
would conservatively bound typical plant piping geometries by applying ASME ANSI-OM3 to 181
different piping setups. The setups covered most of the configurations that one would encounter inside a
plant. All setups had adequate static and seismic designs that met regulatory design rules. 99.7% of the
analyzed geometries displayed allowable velocities above 12 mm/s, which makes the proposed
screening level appear very conservative.
On the other hand, NUREG-1061 considers field experience. It states that for typical power and process
plant piping systems, including appended equipment and supports, the allowable level is approximately
40 mm/s. This is clearly a much higher allowable level than the theoretical value. One reason for the
difference might be the use of linear boundary conditions in the analysis that have a tendency to cause
higher stresses due to the rigidity of the model. Field experience likely provides a better value.
FAI's Nuclear Technical Bulletins - Subscribe Today
Since the age of a plant, operational wear, and general deterioration will have an effect on the fatigue
strength as well, it is suggested to take that into consideration somewhat. Based on the experience at
Fauske and Associates, LLC (FAI), it is recommended to use an allowable peak level of 25 mm/s for
screening purposes.
Based on our field experience, vibration levels below this threshold should cause few, if any, problems.
Vibration levels near this threshold are considered a rough running configuration. Small bore pipes have
a tendency to undergo damage first. That is in particular true for unsupported configurations such as
drains (drip legs) or sampling lines. These lines exhibit little damping and are easily excited. Tieback
supports can provide quick remedy. Safety related equipment that is installed on the pipe line (e.g.
automatic isolation valves) should be evaluated separately and is likely to require a lower allowable
level.

Figure 1 FAI Engineer Measuring Vibration Data on a Steam Valve Stem

In summary, vibration velocity is considered as an adequate general indicator of vibration severity and
distress for piping systems. It can be used as acceptance criterion to assess the severity of piping
vibration. The threshold at which the vibration levels are classified as unsafe must be derived from a
stress analysis report. In the absence of a detailed analysis, it is recommended to use 25 mm/s peak
vibration velocity as screening criterion