Flow measurement

Every device in this article does the same two things. It arranges for a moving fluid to produce a pressure difference that would not exist if the fluid were still, and it hands you an equation, built from continuity and the Bernoulli equation, that converts the pressure back into a speed. The interesting engineering is not in that equation, which is short. It is in how much of it is a lie, and what each instrument does to make the lie small enough to live with.

The Pitot-static tube

Point an open tube directly into a stream and the fluid arriving along its axis has nowhere to go. It decelerates to rest right at the tip, and that point is called a stagnation point. Apply Bernoulli along the streamline that ends there, from a point upstream where the pressure is \(P\) and the speed is the \(V\) you want, to the tip itself where the speed is zero and the pressure has risen to the stagnation value \(P_0\):

Rearranged, \(V = \sqrt{2(P_0-P)/\rho}\). The whole instrument is built to deliver the two pressures on the right of that equation cleanly. A Pitot-static tube reads \(P_0\) through a hole at its very tip and \(P\) through a ring of small holes set back along the shaft, far enough downstream of the tip that the flow past them has recovered to the undisturbed static pressure but not so far that the shaft's own presence has disturbed it again. A manometer or pressure transducer connected across the two reads \(P_0 - P\) directly, and that single number is all the instrument ever measures.

Two things go wrong in practice, and both are about the static port rather than the stagnation one. If the ports sit too close to the tip, they read some of the flow's acceleration around the nose and under-read the true static pressure. If the tube is misaligned with the flow by more than a few degrees, the stagnation reading itself falls off, because the streamline that reaches the tip is no longer the one travelling at the free-stream speed. Weather-vane mounting or a hemispherical tip that is tolerant of yaw are both answers to that second problem, not the first.

The tube gives a velocity at one point, not a flow rate, which is exactly why it is the instrument in a pilot's airspeed indicator and not the meter on a water main. Getting a flow rate out of it means traversing the tube across the duct and integrating a velocity profile by hand, which is fine in a laboratory and hopeless as a permanent installation. The remaining instruments in this article all trade that generality for a single, continuously readable number.

The venturi meter

Instead of stopping the flow, narrow it. A venturi meter is a smoothly converging section of pipe leading to a throat of known, smaller diameter, followed by a gentle diverging section back to the original bore. Continuity between the full-bore inlet, station 1, and the throat, station 2, fixes the throat velocity in terms of the one you want:

Bernoulli between the same two stations, both on the pipe axis so elevation drops out, relates the pressures to the speeds. Substitute the continuity result and solve for the throat velocity:

Multiply by the known throat area \(A_2\) and the result is a flow rate, read continuously off a single pressure difference between two taps that never move. That is the appeal over a Pitot tube: no traverse, no assumption about the shape of the velocity profile, just one number in and one number out.

The diffuser earns its keep after the throat. A sudden expansion back to full bore would dump the kinetic energy the flow picked up in the throat into turbulence and lose it, the way it is lost at the exit of a sharp-edged fitting; a gradual one, at an included angle of somewhere around 7 to 15 degrees, lets the flow decelerate against a rising pressure without separating, and most of that kinetic energy reappears as pressure recovered downstream. The venturi is, in effect, a device for measuring a pressure drop while giving almost all of it back.

That gentle geometry is also the expense. A venturi meter with a proper diffuser is long, several pipe diameters at least, and it has to be cast or machined to a smooth converging-diverging contour rather than bolted together from flat stock. Nothing about the equation above changes if you replace that contour with something cruder; what changes is how honestly the equation describes what actually happens, which is the subject of the next two sections.

Orifice and nozzle meters

A flow nozzle keeps the venturi's smoothly converging throat and throws away the diffuser. Flow accelerates into the nozzle exactly as it did before, so the same continuity and Bernoulli argument gives the same relation between \(V_2\) and \(P_1-P_2\). But the jet now leaves the nozzle directly into the full-bore pipe with nothing to guide its expansion, and that sudden, unguided expansion is turbulent and irreversible. The pressure drop measured across the nozzle is largely not given back. A flow nozzle is shorter and cheaper to manufacture than a venturi, at the cost of turning most of that reading into a permanent loss rather than a temporary one.

An orifice meter goes further still: a thin flat plate with a circular hole machined in it, clamped between two pipe flanges. There is no attempt at a smooth throat at all, and the consequence shows up immediately downstream of the plate. Fluid approaching a sharp-edged hole cannot turn the corner into it; its inertia carries it inward past the edge, and the jet continues to contract for a short distance after the plate before it reaches a minimum area and begins to spread back out. That minimum section is called the vena contracta, and it is smaller than the hole itself.

This matters because the equation was derived assuming the flow area at the measurement station is the known, machined area \(A_0 = \tfrac{\pi}{4}D_0^2\). At the vena contracta it is not; it is some smaller area that depends on the hole's geometry and, weakly, on the flow itself, and it is not something you can look up on a drawing. An orifice plate is the cheapest of the three meters to buy and by far the easiest to install or swap out, which is exactly why it remains the most common flow-measuring device in industry despite being the least accurate of the three.

Whichever of the three throats you have chosen, the equation for the throat velocity in the previous section is unchanged in form. What has changed is how much you should trust the number it returns, and that gap between the equation and reality is what the next section gives a name to.

The discharge coefficient

Bernoulli's equation is for a flow with no friction, and its derivation for a venturi or an orifice quietly assumes the jet at the measurement station exactly fills the known area and moves past it at a single, uniform speed. Real flows fall short of both assumptions: viscous friction along the throat dissipates a little energy that the equation did not budget for, and at a sharp orifice the vena contracta shrinks the effective area well below \(A_0\). Both effects make the actual flow rate smaller than the ideal one computed from the pressure drop alone.

Rather than model either effect from first principles, which is impractical for a machined edge whose exact contour varies plate to plate, the whole shortfall is absorbed into a single empirical multiplier, the discharge coefficient:

\(C_d\) is measured, not derived, tabulated as a function of the diameter ratio \(\beta\) and the Reynolds number, and it is the reason every one of these meters ships with a calibration chart or a certificate rather than a promise that the bare equation will do. A well-made venturi, with its gently guided, nearly loss-free throat, has \(C_d\) close to unity, typically around 0.97 to 0.98. A long-radius flow nozzle sits in much the same range, since its throat is smooth even without a diffuser. A concentric, sharp-edged orifice is a different story: the vena contracta alone accounts for most of the shortfall, and a typical \(C_d\) is around 0.6 to 0.65.

It is worth being precise about what \(C_d\) is not. It is not a fudge factor applied because the engineer was careless with the derivation; the derivation is exact for an inviscid, uniform, one-dimensional flow, and no real flow through a real throat is any of those three things. It is not a constant either, in the strict sense: near the low end of the Reynolds number range it drifts, which is why the calibration is normally quoted alongside the range of flow rates over which it was measured.

Choosing a meter

With the same equation underneath all three devices, the choice between them is an argument about everything the equation leaves out: what the meter costs to buy and install, how much of the measured pressure drop is returned to you as usable head rather than dissipated, as a permanent head loss the pump downstream now has to make up for the life of the installation, and how confidently the discharge coefficient is known.

The venturi sits at one end of every trade-off. It is the most expensive to manufacture and the longest to fit into a pipe run, several diameters of smooth casting rather than a thin plate between two flanges, and it demands the most careful installation to keep its calibration valid. In exchange it recovers most of the pressure drop it borrows through the diffuser, so the permanent head loss it leaves behind is small, and its discharge coefficient is both high and unusually stable across a wide range of flow rates. Where the pumping cost of a permanent loss matters over years of operation, that trade is worth making.

The orifice plate sits at the other end. It is cheap, it is thin enough to swap during a routine shutdown, and it can be resized for a new flow range for the price of a new plate rather than a new meter run. What it gives up is almost the entire pressure drop it creates, since there is no diffuser to hand any of it back, and a discharge coefficient that is lower and more sensitive to the exact sharpness of the machined edge. Where flow rate needs to be read rather than optimised, and the installation is going to be revisited and recalibrated periodically anyway, that is a fair price.

The flow nozzle is the compromise, closer to the venturi's accuracy because it keeps a smooth throat, but without the diffuser's cost and length, and so without its pressure recovery either. It turns up where a venturi's installed length will not fit and an orifice's accuracy will not do, which in practice is a narrower set of jobs than either of the other two.