Liquid Flow in Control Valves – Choked flow, Cavitation

and Flashing
Choked Flow

The basic liquid sizing equation, shown in the upper left of Figure 1, tells us that the
liquid flow rate through a control valve is proportional to the square root of pressure
drop. This simple relationship is shown graphically by the green portion of the graph in
Figure 1. (Note that the scale of the horizontal axis is the square root of pressure drop.)
This linear relationship does not always hold true. As the pressure drop is increased by
lowering the downstream pressure, the flow reaches a point where it no longer
increases. Once this happens, additional increases in pressure drop across the valve do
not result in additional flow, and flow is said to be choked. Here we will call this limiting
or choking pressure drop ΔPchoked, so as to be in agreement with the latest versions of
the IEC and ISA control valve sizing equation standards. (Both standards are technically
identical.) Prior to the issuance of the 2011 edition of the IEC valve sizing equation
standard and the 2012 version of the ISA valve sizing equation standard there was no
official name given to the dividing line between non-choked flow and choked flow, so
valve manufacturers made up their own names. Some of the most common ones are
listed in Figure 1.

Figure 1. Liquid flow in a control valve as a function of the pressure drop across the valve.

Let’s take a look at what is happening inside the valve to cause this choking of the flow.
When the flow stream passes through the vena contracta (the point at which the cross
sectional area of the stream is at a minimum), the flow velocity reaches a maximum.
Conservation of energy dictates that since kinetic energy at the vena contracta has
increased to a maximum, potential energy, in the form of static pressure, must decrease
to a minimum. This is illustrated graphically in the lower left of Figure 2. Note that in the
figure, ΔP is less than ΔPchoked and flow is not choked.

Figure 2. Velocity and pressure profile in a control valve without choked flow.

With a liquid at rest, lowering the surrounding pressure to the vapor pressure of the
liquid will cause the liquid to begin vaporizing. Because any particular group of liquid
molecules is in the vena contracta for a very short time, experimentation has shown that
the pressure at the vena contracta must actually drop slightly below the vapor pressure
of the liquid before vaporization begins. The amount below the vapor pressure to which
the vena contracta pressure must drop for vaporization to begin and flow to choke is
approximated in the IEC and ISA Standards by the Liquid Critical Pressure Ratio Factor,
FF. Figure 2 shows the ISA and IEC equation that is used to calculate F F. So, in the
ISA/IEC equation that calculates the value of ΔP choked, the vapor pressure of the liquid is
multiplied by FF or as it is written in the equation for ΔPchoked, “FF PV.” It should be noted
that very often in text books and manufacturer’s literature the subject of the Liquid
Critical Pressure Ratio Factor is not mentioned, but it is this writer’s opinion that the very
slight increase in complexity here is justified in order to give the reader a fuller
understanding of the subject of liquid flow in control valves.

If flow increases to the point that the vena contracta pressure drops to F F PV, vapor
bubbles form in the vena contracta. Any additional decrease in the downstream
pressure causes more bubbles to form, but the pressure at the vena contracta does not
decrease below FF PV. At this point it is worth noting that the flow through a control valve
depends on the pressure difference between P 1 and Pvc (the pressure at the vena
contracta), and since the vena contracta pressure does not decrease below F F PV, the
flow does not increase, resulting in choked flow. Figure 3 illustrates the choking process
along with the cavitation discussed in the next paragraph. Note that in the figure, ΔP is
greater than ΔPchoked; and flow is choked.

Cavitation

As the bubbles move down stream, the cross-sectional flow area opens up, the velocity
goes down and the pressure goes up. Now we have bubbles with an internal pressure
equal to the vapor pressure surrounded by a higher pressure. The bubbles collapse in
on themselves. This combination of bubble formation and the resulting choked flow,
along with the collapse of the bubbles downstream is called CAVITATION. When the
bubbles collapse they make a popping sound. The result is a noise like gravel going
through the valve. This noise can be loud enough to be very annoying and even loud
enough to damage the hearing of a person who is exposed to it for long periods. Also,
when the bubbles collapse, they create shock waves that can cause severe damage to
the valve. The appearance of cavitation damage is a rough, cinder like, look. (See the
picture of a globe valve plug in the upper right side of Figure 3.) This damage can
happen very quickly, sometimes in as little as a few weeks or months. Because
cavitation damage happens so quickly, we try to avoid cavitation at all costs. Very hard
materials give some improvement, but usually the improved performance is not enough
to justify the cost.

Figure 3. Velocity and pressure profile in a control valve with choked flow and cavitation.
Flashing

If we continue to decrease the downstream pressure, we reach a point where the

Figure 4. Velocity and pressure profile in a control valve with choked flow and flashing.

Now, instead of collapsing, the bubbles become larger and very soon transition from
liquid with bubbles in it to vapor with small drops of liquid in it. This is called FLASHING.
The appearance of flashing damage is quite different from cavitation damage, and
appears as smooth, shiny rivers and valleys. (See the picture of a globe valve plug in
the upper right side of Figure 4.)  The damage mechanism is a sand blasting effect.
Downstream of the vena contracta the flow consists of a large volume of vapor with
many tiny drops of liquid. Because the volume increases greatly when liquid vaporizes,
the downstream velocity can be several hundred feet per second, and the high velocity
liquid droplets can erode away a valve part. The damage caused by flashing does not
usually happen as quickly as that caused by cavitation. The use of hard or erosion
resistant materials can often bring the damage to within tolerable limits. Trim parts
made of the hard stainless steels, such as 17-4 ph, hold up quite well, and 316ss or
chrome moly bodies do much better than carbon steel. The existence of flashing
conditions is dictated by the system (P2 is less than Pv) and the valve selection neither
causes or prevents flashing. The noise caused by flashing is usually below 85 dBA and
to the author’s knowledge there is no method for calculating flashing noise.
The Real Situation

Figure 1 and the associated discussion of liquid choked flow is the classical discussion,
and implies that there is a sudden transition from non-choked flow to fully choked flow.
In reality, at pressure drops approaching, but below the calculated value of ΔP choked,
there is usually some formation of vapor bubbles and some degree of cavitation. Figure
5 shows what really happens as flow transitions from non-choked to fully choked flow.
Figure 5 shows what really happens as flow transitions from non-choked to fully choked
flow.

Figure 5. Actual transition between non-choked and choked flow.

The length of the transition is a function of the shape of the primary restriction in the
valve. Many rotary vales have irregular cross-sectional flow areas which can result in
substantial amounts of choking and cavitation at lower operating pressure drops which
can begin in one localized region and gradually spread to the entire restriction as the
pressure drop across the valve increases and flow becomes fully choked.  The
restriction in most globe valves is quite symmetrical, resulting in a shorter transition. The
current ISA and IEC control valve sizing methods do not include a method of calculating
where the transition from non-choked to fully choked flow begins and ends and only
give formulas for calculating the red and green lines in Figures 1 and 5. Figure 5 also
indicates that noise and cavitation damage can begin even before the flow curve begins
to deviate from a straight line. These first stages of cavitation begin when the average
pressure in the main flow jet at the vena contracta is still above F F PV. At points of
abrupt increase in flow area the streamlines that are attached to the physical
boundaries of the valve can separate and when they do, they form vortices or eddies.
The rotational velocity in the eddies can be high enough that the local pressure inside
an eddy drops below the vapor pressure and vapor bubbles form. As the eddy’s
rotational velocity decreases, the pressure surrounding the vapor bubbles increases
and the bubbles collapse causing both noise and damage.

The value of ΔPchoked is a function of both the process conditions (P 1, the pressure
upstream of the valve and Pv, the vapor pressure of the liquid) and the valve’s internal
geometry represented by the experimentally determined Liquid Pressure Recovery
Factor, FL. Typical values of FL are shown in Figure 6. Note that FL is a function of both
valve style and the percentage of valve opening. Higher values of F L are associated with
valves that have a lower potential for choked flow and cavitation, and smaller values of
FL are associated with valves that have a greater potential for choked flow and
cavitation.

There are several methods of increasing the value of ΔP choked and thus reducing the
potential for cavitation and the associated noise and damage: (1) The value of P 1 can
be increased while keeping ΔP the same by moving the control valve to a location
further upstream, or to a location at a lower elevation. (2) The vapor pressure can be
decreased by installing the valve where the liquid temperature is lower, such as the cool
side of a heat exchanger. (3) A valve style with a higher value of F L can be selected. It is
interesting to note that in general, as the F L increases, so does the price of the valve.
There are special cavitation resistant adaptations of many of the valve styles that have
larger values of FL than those shown in Figure 6, yet which retain the other desirable
features of that style.
Before actually purchasing a control valve, it is always a good idea to ask for the manufacturer’s or the
manufacturer’s representative’s comments regarding your selection.

A more extensive discussion of liquid flow in control valves can be found in Chapter 4 of Valin
Corporation’s book, Control Valve Application Technology

Gas Flow in Control Valves

Figure 1. Flow of gas vs. pressure drop ration through a control valve.

The equation for flow at the top of Figure 1 is almost identical to the equation we would use for
liquid in cases where flow was given in pounds per hour. (Note that the subscript, 1, for
pressure and density indicate that they are the conditions upstream of the valve.) The only
difference is that instead of using the square root of pressure drop (Δp) in the equation and for
the scale of the horizontal axis of the graph, we use the square root of the “Pressure drop ratio,”
Δp/p1.  We then substitute the single character, x, for Δp/p1 to make the expression simpler.
Making this change makes the expression “Δp” which would appear in the equivalent liquid
equation, equal to “x p1.” (Δp/p1 X p1 = Δp) This change from the liquid equation is not
absolutely necessary, but we will see later that it makes the prediction of gas choked flow much
more convenient. The equation at the top of Figure 1 tells us that flow is proportional to the
square root of x. Graphing the equation results in the upward sloping green line.

If we were to conduct a flow test, the actual relationship between flow and pressure drop ratio
would be as shown by the curved blue line, not the straight one. At low pressure drop ratios the
flow follows the straight line, but then it deviates more and more until at last, further increases in
pressure drop ratio do not yield any additional flow. At this point we say that flow has become
choked. Since for gas flow, we have chosen to call the horizontal axis the “x” axis instead of the
delta p axis, we define the pressure drop ratio at which flow becomes fully choked as the
Terminal Pressure Drop Ratio, and give it the symbol xT, T standing for “Terminal.”

Let’s see what is going on inside the valve to cause this choking of the flow and to give the
graph its shape

Figure 2. Gas velocity and pressure profile inside a control valve

At this point, I need to point out that in addition to flow being proportional to the square root of
the pressure drop ratio, it is also proportional to the square root of the density at the vena
contracta.  This is true for both liquid and gas, but with liquids (which are incompressible) we
don’t need to make an issue of the fact, because the density at the vena contracta is exactly the
same as the density upstream of the valve. Also, with liquids, the density at the vena contracta
does not change as the flow rate changes.

The velocity of the gas flowing through a valve reaches a maximum at the vena contracta. Due
to conservation of energy, as a result of the velocity increase, the pressure decreases to a
minimum at the vena contracta. When the pressure decreases the gas becomes less dense.
Since flow is proportional to the square root of density at the vena contracta, the decrease in
density causes the flow to be less than it would be if gas were not compressible, accounting for
flow graph starting to round off instead of following the straight line.

Figure 3. Density change and vena contracta enlargement are responsible for the shape of the flow graph.

As we continue to increase the pressure drop ratio, the velocity at the vena contracta becomes
greater and the pressure becomes less, resulting in an even lower density. Now the flow
deviates even more from the straight line that assumes a constant density at the vena contracta
as would be the case for a liquid.

At some point, as the pressure drop ratio is increased and the flow rate increases, the velocity
at the vena contracta becomes sonic. Because the vena contracta is downstream of the
physical restriction and has a smaller cross sectional area than that of the physical restriction,
even though the velocity has reached the maximum velocity that is possible at a restriction, it is
still possible for the flow rate to increase. As the pressure drop ratio is increased further, the
vena contracta starts backing up toward the physical restriction and the cross sectional area of
the vena contracta increases, so even though flow is sonic there is still some increase in flow
because the area is larger. Finally, when the vena contracta backs up to the physical restriction,
it can get no larger, and since flow is already sonic, no increase in flow is possible, and flow
becomes fully choked.

Summarizing how the gas flow graph gets its shape:

At vena contracta velocities below sonic, the deviation of the flow curve from a straight line is
caused by the density at the vena contracta decreasing. Once sonic velocity is reached, the
velocity, pressure and density at the vena contracta remain constant, but the vena contracta
backs up toward the physical restriction, becoming larger and thus allowing flow to increase.
When the vena contracta finally reaches its maximum size (and since the velocity is already at
the maximum possible) flow chokes.

Now let’s see why we plot flow against pressure drop ratio instead of just pressure drop.

Figure 4. The rated Terminal Pressure Drop Ratop, xT, of an 80% open globe valve.

Let’s see what would happen if we were to run the three flow tests listed in Figure 4, using a
typical GLOBE valve, with the inlet pressure, p1, at first 100 psia, then at 200 psia, and finally at
1,000 psia.

With p1 at 100 psia and starting with delta p at zero and gradually increasing it we would find
that flow would choke when the pressure drop was 70 psi.

Repeating with p1 = 200 psia, flow would choke at a pressure drop of 140 psi.

Finally, repeating the test with p1 = 1000 psia, flow would choke at a pressure drop of 700 psi.
Now, if we calculate the value of xT for each of the tests (since x is delta p divided by p1, xT, the
choked value of x, is delta p Choked / p1) we see something very interesting. xT turns out to be
0.7 in each case.

The point here is that for a particular style of control valve at a particular degree of valve travel
(in this case an 80% open globe valve) the pressure drop ratio at which flow becomes choked is
a constant. Knowing that the terminal pressure drop ratio for this particular model globe valve is
0.7 we can now predict that if the inlet pressure was 300 psi, flow would choke at 210 psi
pressure drop (300 X 0.7 = 210).

In order to properly size control valves for gas service, it is necessary to know at what pressure
drop flow will choke. Different valve styles have different values of xT , and for each valve type,
xT also varies with valve opening.

Figure 5. Typical values of xT and how they affect the flow through control valves.

Valve manufacturers test their valves for xT and then publish the results, making it possible to
predict the point at which flow will choke and therefore properly size control valves.

The blue line on the left hand graph represents the flow through a globe control valve, where xT
is 0.7. (That is, flow will choke when the pressure drop is 70% of p1. As an example, this upper
line could represent a 6 inch globe valve at 80% open which would have a Cv of about 250. A 4
inch high performance butterfly valve operating at around 70% open would also have a Cv of
about 250. Although both valves have the same flow capacity (CV), the graph of flow through
the butterfly valve (red line on the left hand graph) looks quite different than the graph of flow
through the globe valve. That is because it has an xT of 0.4, meaning that flow chokes when the
pressure drop is 40% of p1. At low pressure drop ratios, the flow is the same through both
valves, but as the pressure drop ratio increases, the flow in the butterfly valve starts aiming
toward choked flow before the flow in the globe valve does. Understanding this will help you
understand why a sizing calculation may show that, with all the flow conditions the same, one
style of valve will need a larger Cv than is required of a different style of valve.

Before we conclude by showing how the ISA/IEC control valve sizing equations accurately
predict both the shape of the gas flow vs. pressure drop ratio curve and the point at which flow
chokes, we need to introduce one more concept, that of the “Ratio of Specific Heats Factor,” Fγ
The valve manufacturer’s published values of xT are based on choked flow tests using air as
the test medium. Many gasses other than air have sonic velocities that differ from that of air, so
to compensate for the sonic velocity of those gasses, the published value of xT is multiplied by
the Ratio of Specific Heats Factor, Fγ (F sub gamma), of the gas for which the valve is being
sized. The Ratio of Specific Heats Factor is calculated by dividing the Ratio of Specific Heats, γ,
by the Ratio of Specific Heats of air, which is 1.4.

Fγ = γ/ 1.4

Fγ of air reduces to 1.0.

Most tables of gas properties include values of the Ratio of Specific Heats.
Figure 6. The ISA/IEC gas flow equation including the "Expansion Factor", Y, which compensate for density change at the vena
contracta and vena contracta enlargement.

Note that older versions of the ISA control valve sizing Standard, and some manufacturer’s
literature, use the symbol “k” for the Ratio of Specific Heats.

The equation at the top left of Figure 6 is the ISA/IEC gas equation with mass flow in pounds
per hour as the dependent variable.

Because it is not easy to determine what the density of the gas is at the vena contracta (which
varies with valve style, valve opening and flow rate), the ISA/IEC control valve gas equation
uses the (easy to determine) upstream density (rho sub 1). Recall that earlier, when we
discussed why the actual flow graph has the shape it does, we said that the first part of the
curved portion is the result of changes in density at the vena contracta, but the second portion
(when flow at the vena contracta is sonic and density remains constant) is due to enlargement
of the vena contracta as it backs up to the physical restriction. So, even if we could determine
the density at the vena contracta, that wouldn’t be enough to give the flow graph its correct
shape. To accurately size control valves for gas service, and give the flow graph is correct
shape, an Expansion Factor (symbol Y) is added to the equation we started with earlier, to
correct the calculated flow (and the graph) for both changes in density at the vena contracta and
for vena contracta enlargement. The equation shown here for Y is based on experimental
observation of what actually happens.

Y is a function of x, the pressure drop ratio. When plotted on a square root scale, the graph of Y
looks like the red line labeled “Y” in Figure 6.

Multiplying the straight green line (the flow graph if there was no density change and choking)
by Y results in the actual flow graph (blue line). It is important to limit the value of x used in
sizing or flow calculations to the choked value (Fγ xT), otherwise Y would decrease below 0.67
and the predicted flow, after reaching a maximum value at x = Fγ xT  would then decrease,
which we know is not the case.

Figure 7 summarizes the ISA/IEC gas sizing equations. The two most commonly used forms are
shown.

Figure 7. Summary of the most common forms of the ISA/IEC control valve gas sizing equations.

The top equation is the one we have been using so far, but rearranged to solve for Cv instead of
W. This form is appropriate for gasses and vapors (including steam) where flow is in mass flow
units (pounds per hour) and where the upstream density is known. The second equation, which
is simply the first equation, with appropriate unit conversions for flow in volumetric units (scfh)
and with the density calculated from the specific gravity, absolute temperature, absolute
pressure and compressibility factor.