Orifice Plate Noise Is Cavitation, Not Excess Flow Rate

A 5 mm bore letting 5 barg water down to an atmospheric return dissipates its energy in a jet whose core pressure falls below the vapour pressure of the water. The sound is vapour cavities imploding a few diameters downstream, not turbulence in the bore. That distinction decides the fix: the differential across the plate sets the regime, and the flow rate only sets how much acoustic power that regime radiates.

Noise Source: Cavitation at the Vena Contracta

A thin sharp-edged plate is a high-recovery device. The jet contracts to roughly 60 % of the bore area, static pressure at the vena contracta falls far below the downstream pressure, and then most of it recovers. Vapour forms in the low-pressure core and collapses during that recovery. Work the cavitation index for the as-built case:

sigma = (P1 - Pv) / (P1 - P2)

P1 = 6.0 bar a   (5 barg supply, measured at the plate)
P2 = 1.0 bar a   (return outlet at atmosphere)
Pv = 0.023 bar a (water at 20 C)

sigma = (6.0 - 0.023) / (6.0 - 1.0) = 1.20

Conventions differ between sizing tools; some define sigma on downstream pressure. Match the definition your program uses before comparing against its incipient-cavitation limit. At 4 barg supply the number is barely better, and if the same detail is ever reused on hot water the margin disappears entirely: Pv rises to about 0.47 bar a at 80 C, and at 100 C into an atmospheric return the stream flashes rather than cavitates.

Flow rate does not appear in that expression. Dropping the duty from 500 kg/hr to 20 kg/hr cuts the dissipated stream power from about 55 W to 2.2 W (W = dP x Q), which is roughly 14 dB less radiated power, so the line does get quieter. The regime is unchanged. The plate still cavitates, still erodes, and the erosion now attacks a sub-millimetre bore where 0.05 mm of lost edge raises the passing area by a third.

Letdown Options Compared

Multi-hole plates work by moving the characteristic jet frequency up with the hole diameter: smaller holes push energy toward and above the audible band, where pipe-wall transmission loss and A-weighting both work in your favour. They do not raise sigma on their own. Vendors including Spirax Sarco supply modulating freeze-protection valves that throttle on the temperature of the water leaving the trap, which is the only option on the list that actually conserves water.

Recommended Scheme: Back-Pressure First, Then Two Deliberate Stages

Two plates were not madness on the last plant. They were the correct device, fitted without a designed split. Repeat the arrangement, but pick the interstage pressure instead of letting the bores pick it for you.

Take the free pressure first. Elevating the return outlet by 5 m adds about 0.49 bar of back pressure and lifts single-stage sigma from 1.20 to 1.33; 10 m gives 1.49. Useful, nearly free, and almost never enough on its own against a 4-5 bar drop, so budget for the second stage regardless.

Split the remaining drop by equal pressure ratio, r = (P3/P1)^(1/n). For 6.0 bar a to 1.0 bar a in two stages, r = 0.408 and the interstage pressure is 2.45 bar a (1.45 barg). Equal ratio gives equal sigma, so neither stage is the weak one:

The downstream plate is the larger of the two. If it is not, the split has been built backwards and stage 1 will pass its bore wide open while the last plate takes the whole drop.

Sizing Procedure

1. Fix the design flow from the freeze-protection heat balance, not from habit: m = Q_loss / (cp x (T_in - T_out,min)). Use the coldest design ambient, the actual insulation condition and wind exposure.
2. Establish P1 at the plate face and P2 at the plate outlet, including downstream line losses and the static head to the return outlet. Header design pressures are not inputs.
3. Compute sigma. If it falls below the incipient-cavitation limit of your sizing program, stage it; do not reach for a bigger plate.
4. Split by equal pressure ratio across n stages and tabulate the interstage pressures.
5. Size each bore from A = m_dot / (Cd x sqrt(2 x rho x dP)), then d = sqrt(4A/pi). With Cd = 0.6, rho = 1000 kg/m3 and 4 bar across a single plate this gives 0.65 mm at 20 kg/hr, 1.44 mm at 100 kg/hr and 3.2 mm at 500 kg/hr. The installed 5 mm and 6 mm bores pass roughly 1200 and 1730 kg/hr on the same basis, so confirm what the existing plates are really flowing before repeating them.
6. Separate the plates by straight pipe. Allow at least about six pipe diameters so static pressure recovers between stages, and fit an interstage pressure tap. Plates stacked in one flange set behave as a single restriction and the staging does nothing.
7. Specify stainless plates with adequate thickness, and put a strainer upstream sized on the smallest bore in the train.

The 20 kg/hr Trap: Bore Size Against Plugging

Run the heat balance before committing to the reduction. Thirty metres of insulated DN25 losing 10 W/m at design ambient needs 300 / (4190 x 15) = 0.0048 kg/s, about 17 kg/hr, to arrive at 5 C from a 20 C supply. So 20 kg/hr is the right order for a short, genuinely insulated run and nowhere near enough for a bare or wind-exposed one. At that rate the velocity in DN25 is about 0.01 m/s: the line is thermally stratified, the calculated figure is a floor rather than a design point, and dead legs off the bypass are not protected at all.

The device is the harder problem. A 0.65 mm single hole plugs on scale, weld spatter or biofilm, and it fails silently: nothing alarms, and the first indication is a burst line in February. A perforated plate at that duty needs holes finer still. If saving water across the project is the objective, buy the temperature-modulating freeze-protection valve with an upstream filter and install it at the location the manufacturer specifies, and put the staged plates upstream of it so its trim never sees the full 4-5 bar.

Commissioning Checks and Recurring Traps

1. Gauge the interstage tap at design flow. It should read close to the calculated 1.45 barg. Near-atmospheric means stage 2 is passing everything or stage 1 is plugged; near supply pressure means the bores are fitted in the wrong order.
2. Listen and characterise. A steady broadband hiss is flow noise; a rattle like gravel in the pipe is collapsing vapour. Confirm it by throttling a valve downstream of the last plate: cavitation noise drops sharply as back pressure rises, flow noise does not. Use that only as a test, never as the permanent second stage, or the valve trim becomes the sacrificial component.
3. Hand-check pipe vibration at the spool downstream of the last plate. Persistent buzzing there points at residual cavitation, not at supports.
4. Log the freeze-protection outlet temperature at the coldest ambient of the first winter and compare it with the heat-balance prediction. Trend the strainer differential over the same period.
5. At the first shutdown after that winter, pull the spool downstream of the last plate and inspect the bore edges and the pipe wall for pitting; measure the bores and confirm they have not opened up.

FAQ