Venturi Meters, Orifice Plates, and Pitot Tubes

Squeeze the flow or stop it, measure the pressure change, and you've measured the flow.

A 12 in. water main runs under the street, and the utility wants to know how much water it carries. You can't see inside it, and you can't hold a bucket under it. But you can pinch it.

Narrow it to 6 in. for a short stretch, and the water there has to go four times as fast, paying for the speed out of its pressure. A gauge just before the pinch and one in it differ by 5 psi? Then about 5.4 cfs is going by, some 2,400 gallons a minute.

Squeeze the flow: the venturi meter

The pinch is a venturi meter: a short cone narrowing the pipe to a throat, then a longer, gentler cone widening it back. A tap at the inlet, section 1, and another at the throat, section 2, read the pressure where the water is slow and where it's fast.

1 2 flow D1 D2 P/γ + z Δh permanent loss
A venturi meter, with piezometers at the inlet and throat
The dashed line is the level water would reach in piezometers all along the pipe, the piezometric head P/γ+zP/\gamma + z. The small gap at the right, the permanent loss, comes up later.

Pick those two points. Continuity gives v1=v2A2/A1v_1 = v_2A_2/A_1, and Bernoulli, with each section's pressure and elevation heads grouped into its piezometric head from Manometers, gives

(P1γ+z1)−(P2γ+z2)=v222g−v122g\left(\frac{P_1}{\gamma} + z_1\right) - \left(\frac{P_2}{\gamma} + z_2\right) = \frac{v_2^2}{2g} - \frac{v_1^2}{2g}

Call the left side Δh\Delta h, the drop the piezometers show. It pays only for the water's gain in velocity head, since the water arrives already moving. By continuity, the inlet's velocity head is (A2/A1)2(A_2/A_1)^2 times the throat's, so

Δh=v222g[1−(A2A1)2]⇒v2=2g Δh1−(A2/A1)2\Delta h = \frac{v_2^2}{2g}\left[1 - \left(\frac{A_2}{A_1}\right)^2\right] \quad\Rightarrow\quad v_2 = \sqrt{\frac{2g\,\Delta h}{1 - (A_2/A_1)^2}}

The bracket is the approach-velocity term. With a throat half the pipe's diameter, the inlet already has 1/16 of the throat's velocity head. Leave the term out, and the flow comes out about 3% low, and more with a wider throat. Since areas go as the diameter squared, (A2/A1)2=(D2/D1)4(A_2/A_1)^2 = (D_2/D_1)^4, a fourth power.

Multiply by the throat's area for the flow, and by a coefficient of velocity, CvC_v, for the little energy friction takes between the taps, as for a jet leaving a tank in Tanks, Siphons, and Weirs. That's the Handbook's Venturi Meters equation:

Q=CvA21−(A2/A1)22g(P1γ+z1−P2γ−z2)Q = \frac{C_v A_2}{\sqrt{1 - (A_2/A_1)^2}}\sqrt{2g\left(\frac{P_1}{\gamma} + z_1 - \frac{P_2}{\gamma} - z_2\right)}

Here z1z_1 is the entrance and z2z_2 the throat. In a level meter read by gauges, the root holds (P1−P2)/γ(P_1 - P_2)/\gamma, the gauges' difference as a head of the flowing water. Problems give CvC_v.

Worked Example The Hook's meter
1 of 2

A level 12 in. water main has a venturi meter with a 6 in. throat and Cv=0.98C_v = 0.98. Gauges at the inlet and the throat differ by 5.0 psi. Find the flow rate.

Reading the differential on a manometer

The root of the venturi equation holds exactly what a differential manometer reads: the piezometric-head difference. Multiply a mercury manometer's deflection by 12.6, not 13.6, since water fills the rest of both legs, and you have Δh\Delta h in feet or meters of water.

That holds at any slope. Tilt the meter or stand it on end, and the taps' difference in elevation cancels inside the manometer, as on the sloping pipe in Manometers. Pressure gauges are different: on a meter that isn't level, you add z1−z2z_1 - z_2 yourself.

Checkpoint
1 of 2

A venturi meter in a level water line has a 6 in. inlet, a 3 in. throat, and Cv=0.98C_v = 0.98. A mercury manometer across its taps reads 7.5 in. What is the flow rate?

cfs

The orifice plate: a cheaper squeeze

A venturi is a long, machined casting, priced like one. The cheap alternative is a thin plate with a sharp-edged hole, bolted between two flanges. As at a sharp-edged hole in a tank, the water can't turn the corner at the edge, so the jet keeps narrowing past the plate to a vena contracta, with dead water swirling around it.

1 2 flow A1 A0 A2 vena contracta P/γ + z Δh permanent loss
A sharp-edged orifice plate, with piezometers upstream and at the vena contracta

Adapted from the FE Reference Handbook, Fluid Flow Measurement, Orifices

The downstream tap goes at the vena contracta, the narrowest, fastest, lowest-pressure section, so the vena contracta plays the throat. Its area is A2=CcA0A_2 = C_cA_0, with A0A_0 the hole's area. Put CcA0C_cA_0 in for A2A_2 in the venturi equation, and out comes the Handbook's Orifices equation:

Q=CA02g(P1γ+z1−P2γ−z2)C=CvCc1−Cc2(A0/A1)2Q = CA_0\sqrt{2g\left(\frac{P_1}{\gamma} + z_1 - \frac{P_2}{\gamma} - z_2\right)} \qquad C = \frac{C_vC_c}{\sqrt{1 - C_c^2(A_0/A_1)^2}}

The Handbook calls CC the coefficient of the meter: the jet's corrections, CcCvC_cC_v, with the approach-velocity term folded in. It multiplies A0A_0, the hole's area, not the pipe's. For a level meter, the Handbook also writes it as Q=CA0(2/ρ)(P1−P2)Q = CA_0\sqrt{(2/\rho)(P_1 - P_2)}.

For a sharp-edged opening, the Handbook's Orifices and Their Nominal Coefficients table gives Cc=0.62C_c = 0.62, Cv=0.98C_v = 0.98, and C=0.61C = 0.61. That 0.61 is just CcCvC_cC_v, with no approach-velocity term, as suits a hole in a tank wall. In a pipe, the term grows with the hole: a 3 in. hole in a 12 in. pipe leaves CC at 0.61, but a 6 in. hole in an 8 in. pipe raises it to 0.648. Use 0.61 when a problem calls for the nominal coefficient, and build CC when it gives CcC_c, CvC_v, and the area ratio.

Worked Example An orifice plate read by mercury
1 of 2

A 250 mm water line has a sharp-edged orifice plate with a 150 mm hole, Cc=0.62C_c = 0.62, and Cv=0.98C_v = 0.98. A mercury manometer across its taps reads 120 mm. Find the flow rate.

Running a meter backward

A meter's equation runs backward as easily as forward. For the differential a venturi will show at a required flow, solve for what sits under the root:

Δh=[1−(A2A1)2]12g(QCvA2)2\Delta h = \left[1 - \left(\frac{A_2}{A_1}\right)^2\right]\frac{1}{2g}\left(\frac{Q}{C_vA_2}\right)^2

For an orifice plate, it's Δh=(Q/CA0)2/2g\Delta h = (Q/CA_0)^2/2g. Then turn Δh\Delta h into whatever the instrument reads: multiply by γ\gamma for a pressure difference, or divide by 12.6 for a mercury deflection on a water line.

The throat diameter is less direct, because D2D_2 sits in both A2A_2 and the approach-velocity term. Go back to Bernoulli instead. With the ideal flow Q/CvQ/C_v, what the same differential would drive without friction, the throat's velocity head is the inlet's plus the differential:

v1=Q/CvA1v222g=v122g+ΔhA2=Q/Cvv2v_1 = \frac{Q/C_v}{A_1} \qquad \frac{v_2^2}{2g} = \frac{v_1^2}{2g} + \Delta h \qquad A_2 = \frac{Q/C_v}{v_2}

Or hand the meter equation to your calculator's equation solver.

Checkpoint
1 of 2

A venturi meter with a 300 mm inlet, a 150 mm throat, and Cv=0.98C_v = 0.98 will measure a water flow of 0.10 m³/s. What deflection will a mercury manometer across its taps show?

mm

What a meter costs to run

In both figures, the piezometric head bottoms out where the water moves fastest. Nowhere else in a meter is the pressure as low.

Past that point, the water slows back down, and how gently it does so sets what the meter costs to run. The venturi's long cone widens so gradually that the water follows the walls and turns most of its extra velocity head back into pressure. Past an orifice plate, the jet shoots into slower water, and the eddies churn most of it into heat. From the same Δh\Delta h, the venturi gets most back and the orifice plate loses most for good. That loss is the meter's permanent head loss, and the pumps pay for it as long as the meter stays in the line. The Handbook prints no values for it, so here it's only compared; The Energy Equation puts head losses into numbers.

The coefficients describe only the stretch between the taps. A venturi's inlet cone narrows gently, so the water follows its walls and fills the throat: no vena contracta, just a little friction to correct for, and CvC_v is near 1, often about 0.98. An orifice's jet shrinks to 62% of the hole, so its CC sits near 0.61. The plate's virtue is its price.

In both meters, Δh\Delta h pays for a velocity head, so it grows with the square of the flow: the Hook's meter at twice the flow would show 20 psi, not 10.

Checkpoint

Which statements about venturi meters and orifice plates are true?

A.
If a venturi meter's differential rises from 2 psi to 8 psi, the flow through it has quadrupled.
B.
Cutting off a venturi's long downstream cone would leave its coefficient near 1 but raise its permanent head loss.
C.
In an orifice-plate meter, the pressure is lowest just upstream of the plate, where the water presses against it.
D.
Swapping a venturi for an orifice plate that shows the same differential at the same flow would lower the pumping cost.
E.
Halving the flow through an orifice plate cuts its differential to a quarter.

Stop the flow: the pitot tube

Squeezing is one way to make a flow trade pressure for speed. The other is to stop it. Point a thin tube into the flow, open end upstream. The water arriving at the opening has nowhere to go, so it comes to rest at the tip, the stagnation point, and its velocity head turns into pressure.

Pick two points on the streamline into the tip: one just upstream, moving at vv with the static pressure PsP_s, and one at the tip, at rest, with the stagnation pressure P0P_0. At one elevation, Bernoulli leaves

P0γ=Psγ+v22g\frac{P_0}{\gamma} = \frac{P_s}{\gamma} + \frac{v^2}{2g}

The stagnation pressure beats the static by one velocity head. Solved for vv, that's the Handbook's Pitot Tubes equation:

v=2ρ(P0−Ps)=2g(P0−Ps)γv = \sqrt{\frac{2}{\rho}(P_0 - P_s)} = \sqrt{\frac{2g(P_0 - P_s)}{\gamma}}

In an open stream, the free surface shows the static pressure, and the water in a bent tube facing the current climbs one velocity head above it:

P0 v, Ps Ps/γ v2/2g
A pitot tube in an open stream

Adapted from the FE Reference Handbook, Fluid Flow Measurement, Pitot Tubes

A pipe has no free surface, so a pitot-static tube packs both openings into one probe: a nose hole for P0P_0, and side holes, which the flow slides past without stopping, for PsP_s. Read the manometer across them as in Manometers, with the flowing liquid as the line liquid:

flow P0 Ps water mercury h
A pitot-static tube on a mercury manometer
  • An inverted U-tube with air above the flowing liquid: the liquid stands higher on the stagnation side by exactly v2/2gv^2/2g.
  • A U-tube with a heavier gauge liquid: v2/2g=h(SGm/SGw−1)v^2/2g = h(SG_m/SG_w - 1), with SGwSG_w the flowing liquid's specific gravity. For mercury on a water line, that's 12.6h12.6h.

In air, keep the two fluids straight: the manometer liquid sets the pressure difference, P0−Ps=γmhP_0 - P_s = \gamma_m h, and the air's density sets the velocity. The Handbook's Fluid Mechanics chapter has no air table, but its Selected Properties of Air, under Environmental Engineering, gives 0.0734 lbm/ft³ at 80°F, or 0.0734/32.2=0.002280.0734/32.2 = 0.00228 slug/ft³. In SI, expect the density to be given, about 1.2 kg/m³ at room temperature. The equation assumes an incompressible fluid, which the Handbook accepts for a gas up to Mach 0.3, about 100 m/s in room-temperature air.

Worked Example A pitot-static tube in an air duct
1 of 4

A pitot-static tube in a duct carrying 80°F air is connected to a U-tube of water, which reads 0.60 in. Find the air's velocity at the tube's tip.

What a pitot tube reads

A pitot tube reports the velocity at its tip and nowhere else. Water in a pipe moves fastest on the centerline and hardly at all at the wall, while Q=AvQ = Av needs the average across the section. And the reading is a velocity head, so it goes as v2v^2: nine times the reading means three times the speed.

Checkpoint

A pitot-static tube sits on the centerline of a water pipe. Which statements are true?

A.
Its velocity times the pipe's area overestimates the flow.
B.
If the velocity doubles, its manometer's deflection doubles.
C.
Moved close to the pipe wall, it would read less than the pipe's average velocity.
D.
Its side holes face the flow head-on, so they sense the stagnation pressure.
E.
If its manometer's deflection quadruples, the velocity at the tip has doubled.

Recap

  • Squeeze a flow, and its pressure drops to pay for the speed: Q=CvA22g Δh/1−(A2/A1)2Q = C_vA_2\sqrt{2g\,\Delta h}/\sqrt{1 - (A_2/A_1)^2}, with (A2/A1)2=(D2/D1)4(A_2/A_1)^2 = (D_2/D_1)^4.
  • A differential manometer reads Δh\Delta h at any slope: 12.6h12.6h for mercury on water.
  • An orifice plate's vena contracta is its throat: Q=CA02g ΔhQ = CA_0\sqrt{2g\,\Delta h}, with CC built from CcC_c, CvC_v, and A0/A1A_0/A_1, or the nominal 0.61.
  • Δh\Delta h goes as Q2Q^2. A venturi's CvC_v is near 1, and it recovers most of Δh\Delta h; an orifice plate loses most of it.
  • A pitot tube reads one point's velocity, v=2(P0−Ps)/ρv = \sqrt{2(P_0 - P_s)/\rho}, with the flowing fluid's ρ\rho.