HARMONIC ANALYSIS
Harmonic analysis looks at dynamic problems where the forces or displacements acting
on the piping system take sinusoidal forms. The dynamic motion equation could be
solved directly thus obviating the need for response spectrum.
Loads with harmonic force/time profiles are best solved using the harmonic
method. Major types of loads with harmonic time profiles are:
• Acoustic vibration: If fluid flow characteristics are changed within a pipe (for
example if flow conditions change from laminar to turbulent as the fluid goes
through an orifice), slight lateral vibrations may be set up within the pipe. Often
these vibrations fit harmonic patterns, with predominant frequencies somewhat
predictable based upon the flow conditions. For example, Strouhal’s equation
predicts that the developed frequency (Hz) of vibration caused by flow through an
orifice will be somewhere between 0.2 V/D and 0.3 V/D, where V is the fluid
velocity and D is the diameter of the orifice. Wind flow around a pipe sets up
lateral displacements as well (a phenomenon known as vortex shedding), with an
exciting frequency in the area of 0.18 V/D, where V is the wind velocity and D is
the outer diameter of the pipe.
The biggest use by far of the harmonic solver is in analyzing low frequency field
vibrations. The approach typically taken towards solving this type of problem (especially
if the problem investigated is for an actual existing pipeline where large vibrations are
noticed in operation) is described briefly below:
1. A potential dynamic problem is first identified in the field either in terms of large
cycle vibrations or high stresses (fatigue failure) being present in an existing
piping system, raising questions of whether this represents a dangerous situation.
As many symptoms of the problem (quantifiable displacements, overstress points,
etc.) are identified as possible, for future use in refining the dynamic model.
2. A model of the piping system is built using CAESAR II. This should be done as
accurately as possible, since the system as well as load characteristics affect the
magnitude of the developed response.
3. Postulate the cause of the load and from that, estimate the frequency, magnitude,
point, and direction of the load. This is somewhat difficult because the dynamic
loads can come from many sources. Dynamic loads may be due to internal
pressure pulses, external vibration, flow shedding at intersections, two phase flow,
etc., but in almost all cases, there is some frequency content of the excitation that
corresponds to (and therefore excites) a system mechanical natural frequency. If
the load is caused by equipment, then the forcing frequency is probably some
multiple of the operating frequency; if the load is due to acoustic flow problems,
then the forcing frequency can be estimated through the use of fluid dynamics
equations. Using the best assumptions available, the user should estimate the
magnitudes and points of application of the dynamic load. Note that the point of
application is not necessarily a point of high system response.
6. Once the guilty mode has been identified, it must be eliminated. This is done most
easily by adding a restraint at a high point of the mode shape. If this cannot be
done, the mode may also be altered by changing the mass distribution of the
system (for example, by adding operators to valves). If no modification of the
system is possible, it may be possible to alter the forcing frequency of the load. If
the dynamic load was postulated to be due to internal acoustics, it is
recommended that the pipe not be rerouted at this point, as rerouting the pipe will
change the internal flow conditions. After modifying the system, the harmonic
problem (using the single forcing frequency determined as a good model) is then
re-run, and the stresses, displacements, etc. are re-evaluated.
7. If the dynamic problem has been adequately solved, the system is now reanalyzed
statically to determine the effects of any modifications on the static loading cases.
(Remember that adding restraint normally increases expansion stresses, while
adding mass increases sustained stresses).
The harmonic loading can start with its maximum load at time equal to zero, or
the harmonic load can start with its maximum at any time between zero and t = 2π / ω
Phasing can be important if more than one force or displacement is included. The
phase angle relates the timing of one load to another. For example, if two harmonic loads
are acting along the same line but at different nodes, the loads can be directed towards
each other, which would produce no net dynamic imbalance on the system, or the loads
could be directed in the same direction, which would produce a net dynamic imbalance in
the system equal to the sum of the two forces. It is the phase angle which primarily
determines this relationship. For example, the harmonic load data (1500, X, 0, 10), and
(1500, X, 0, 105) produces an “in phase,” or same direction dynamic load in the system
(1500 in the X direction and zero phase at nodes 10 and 105), while (1500, X, 0, 10), and
(1500, X, 180, 105) produces an “out of phase,” or opposite direction dynamic load on
the system which will tend to pull the system apart.
The two most common phased loadings are those due to rotating equipment and
reciprocating pumps.
Rotating equipment may have an eccentricity, a speed, and a mass. These items
must be converted into a harmonic load that acts on the rotor at the theoretical mass
centerline. The magnitude of the harmonic load (Fh) is computed from:
Fh = (mass) (speed)2(eccentricity)
Where,
Speed: is the angular velocity of the shaft in cycles per second.
This load is applied along both axes perpendicular to the shaft axis and at a 90º phase
shift.
In the case of a reciprocating pump, the pump introduces a pressure wave into the
line at some regular interval that is related to the valving inside the pump and the pump
speed. This pressure wave moves away from the pump at the speed of sound in the fluid.
These pressure waves will cause loads at each bend in the piping system. The load on
each subsequent elbow in the system starting from the first elbow will be phase shifted by
an amount that is a function of the distance between the elbows, from the first elbow to
the current elbow. It is the amount of phase shift between elbow-elbow pairs that
produces the net unbalanced dynamic load in the piping. The phase shift, in degrees from
the first elbow, is calculated from:
Phase = [(frequency)(length) / (speed of sound)]360º
Where,
2. Open the dynamic input processor and select the type of analysis to be
Harmonic.
3. The first tab in the harmonic module is the Excitation Frequencies tab. In
this tab, the starting frequency (Hz.) is entered. Any number of individual
frequencies, or frequency ranges (indicated by a starting, ending, and
incremental frequency) may be specified, one to a line. CAESAR II performs
a separate analysis for each frequency requested. If the ending frequency is
not specified, it defaults to the starting frequency. If the frequency increment
is not specified then it defaults to 1 Hz. The number of anticipated load cycles
may be entered for each frequency range. If the number is entered, the load
cases are calculated with a fatigue stress type. Otherwise, the load cases are
calculated with an occasional stress type.