Why Metal Grilles Whistle in the Wind: Vortex-Shedding, Lock-In Resonance and How to Design It Out
"Fine on a calm day, but it whistles whenever the wind picks up to force 4 or 5" - that is the classic complaint on metal grilles, sun louvres and screen walls. The owner hears a piercing high tone; the contractor tightens every bolt and the whistle comes back with the next gust. Nine cases out of ten this is not an assembly fault - it is fluid mechanics. Air passing a regular section sheds vortices alternately from both sides; when the shedding frequency lands on the bar's natural frequency the two lock together, and no amount of bolt torque will stop it.
The good news: it can be designed out on the drawing board. Below is a three-step quantitative check (shedding frequency, natural frequency, lock-in risk), the two tone sources most people miss, and six measures you can put straight into the shop drawings.
Step 1: Separate the three kinds of "wind noise" - the fixes are different
Site teams call every wind tone a "whistle", but the mechanisms differ and the wrong remedy just burns money:
| Type | What you hear | Mechanism | Direction of fix |
|---|---|---|---|
| Vortex shedding tone | Single steady high tone; appears above a wind threshold and rises in pitch with wind speed | Alternating vortex shedding (Karman vortex street) behind a round or square bar generates an alternating transverse force | Change section / add turbulence breakers / raise natural frequency |
| Slot jet tone | Whistle or organ-pipe character; pitch changes with gap width | Air forced through a 2-8 mm gap accelerates into a jet; edge tone forms at the lip | Change gap width - either open it past 15 mm or close it fully; never leave an awkward middle gap |
| Cavity resonance | Low hum; the whole grille seems to sound; hard to locate | A closed volume between grille and backing plate or column acts as a Helmholtz resonator and amplifies the noise | Add pressure-relief openings, acoustic infill, or a perforated lining inside the cavity |
Do not invert the diagnosis order. First decide whether a single bar is singing or the whole panel is. Single bar, tone rising smoothly with wind speed = vortex shedding. Whole panel, fixed pitch = cavity resonance. Sound only on one wind direction = usually the wake from an opposing building or canopy.
Step 2: Calculate the vortex-shedding frequency
For round and square bars in the normal wind range, estimate the shedding frequency with the Strouhal number:
fs is the shedding frequency (Hz), V the wind speed (m/s) and D the characteristic frontal dimension (m). Put typical grille bar sizes into it and you can see why grilles are so exposed:
| Frontal dimension D | At 5 m/s | At 10 m/s | At 15 m/s |
|---|---|---|---|
| 20 mm tube / 20 mm square | 50 Hz | 100 Hz | 150 Hz |
| 30 mm tube / 30 mm square | 33 Hz | 67 Hz | 100 Hz |
| 50 mm tube / 50 mm square | 20 Hz | 40 Hz | 60 Hz |
| 80 mm tube / 80 mm square | 12.5 Hz | 25 Hz | 37.5 Hz |
| 100 x 20 flat (using 20 mm) | 50 Hz | 100 Hz | 150 Hz |
The ear is most sensitive between 500 and 2000 Hz, but a pure tone anywhere in 20-200 Hz is still clearly heard as a "whistle" - and virtually every conventional bar size lands in that band. That is why grille whistling always generates a complaint: it is not broadband wind noise, it is a tonal whistle that is especially obvious at night.
Step 3: Estimate the natural frequency and check for lock-in
Shedding alone is harmless; the question is whether it meets the bar's natural frequency. Treat a grille bar approximately as a simply supported beam:
The relationship worth memorising: natural frequency is inversely proportional to the square of the span - halve the span and you quadruple the frequency. This is the cheapest and most effective lever available at drawing stage:
| Bar | Span | Natural frequency (order) | Dangerous wind band (fs within 0.8-1.2 fn) |
|---|---|---|---|
| 50 x 2.0 mm tube | 3000 mm | approx. 4-6 Hz | 1.0-1.5 m/s (below threshold - safe) |
| 50 x 2.0 mm tube | 1500 mm | approx. 16-24 Hz | 4-6 m/s (typical wind - dangerous) |
| 50 x 2.0 mm tube | 800 mm | approx. 55-85 Hz | 14-21 m/s (typhoon zone - dangerous) |
| 20 x 1.5 mm tube | 1200 mm | approx. 30-45 Hz | 3-5.4 m/s (worst case) |
Watch for lock-in. When the shedding frequency comes within about 20% of the natural frequency, the vortex shedding is captured by the structural motion: the tone locks onto the natural frequency, stops tracking wind speed, and amplitude rises sharply. This is why the site report is always "fixed pitch, and louder the harder it blows" rather than a tone that slides with the wind - a fixed pitch is essentially confirmation of lock-in.
Step 4: Six design-stage countermeasures
- Break the equal-spacing rhythm (first choice, almost free) - identical spacing, identical diameter and identical length is the worst case. Cycle the pitch, e.g. 100/110/95/105 mm, or insert one different-diameter bar every 6-8 bars. Visually indistinguishable, but the periodic excitation is destroyed.
- Shorten the span or add intermediate supports - taking 1500 mm down to 900 mm lifts the natural frequency by roughly three times and pushes the resonance wind speed past 10 m/s, which removes the "whistles at force 4" complaint immediately. The cost is extra sub-framing - allow for it in the shop drawings.
- Change the section: round to flat-oval, or add longitudinal ribs - a circular section sheds most strongly. For a flat bar, note the short face governs (a 20 mm face sheds at five times the frequency of a 100 mm face). Elliptical and flat-oval sections shed less. Helical strakes - the detail used on stay cables, with a pitch of 5-8 diameters - break up regular shedding and are the usual retrofit tool on existing jobs.
- Make the gap either large or non-existent - a grille gap of 2-8 mm is the high-risk band. Open it past 15 mm so air passes freely, or close it fully with a backing strip. Never leave a half-open gap.
- Add damping at the connections - insert EPDM or polyurethane pads between clips, clamps and bars so vibration energy is dissipated as heat. Specify a weather-resistant elastomer (EPDM or neoprene); ordinary rubber hardens within about two years and then becomes a new source of looseness.
- Spoil the backing cavity or vent it - if there is a closed volume behind the grille, line it with 50 mm glass wool behind a perforated sheet of at least 20% open area. Where infill is impossible, at least drill pressure-relief openings to break the resonance condition. Note that glass wool must be non-combustible and faced for exterior use per local fire code.
Step 5: Loose connections are accomplices, not the culprit
Prolonged whistling accelerates connection failure, and the two feed each other:
- Preload loss - sustained alternating load causes micro-slip in the threads; preload can fall below half the initial value within months. Stiffness drops, the natural frequency drops, and the resonance band moves into normal wind speeds - the whistle threshold falls further.
- Fretting wear - repeated micro-slip rubs through the zinc coating, exposing bare steel; the corrosion product then jacks the contact faces apart, closing the loop.
- Dissimilar metal contact - aluminium grilles bolted directly to stainless or galvanised steel brackets suffer galvanic corrosion wherever a water film forms: the aluminium corrodes faster and the joint works loose.
Anti-loosening practice, in order of value for money:
- Torque to a table, not by feel - a stainless M8 bolt is typically around 15-20 N.m; use the fastener supplier's figures.
- Use nylon-insert lock nuts or a medium-strength threadlocker instead of plain spring washers (spring washers are far less effective against vibration than commonly assumed).
- Double nuts or a wedge-locking washer on critical connections.
- Insulating washers plus nylon sleeves at dissimilar-metal interfaces to break the galvanic path.
- Re-torque every 24 months, with a dedicated check before typhoon season.
Field retrofit sequence: start with the cheapest
On a completed project you cannot start again. Work down this ladder - the whistle usually disappears by step two or three:
| Step | Action | Cost | Success likelihood |
|---|---|---|---|
| 1 | Measure: record with a phone spectrum app and correlate pitch with wind speed | ≈ 0 | - (diagnosis only) |
| 2 | Re-torque every connection to spec and fit EPDM damping pads | Low | About 30% (slot tones and mild resonance) |
| 3 | Fit helical strakes or surface bumps to selected bars to break regular shedding | Medium | About 60% |
| 4 | Add intermediate supports to shorten the span | Medium-high | About 85% |
| 5 | Replace with mixed-diameter or shaped bars to break the panel periodicity | High | About 95% |
Design guidance by product type
| Product | Main risk | Recommended practice |
|---|---|---|
| Aerofoil sun louvres | Section is already foil-like, so shedding is weaker; problems usually come from gaps | Keep the blade lap either under 1 mm or over 15 mm; fit end caps to suppress tip vortices |
| Round-bar grilles / screen walls | Strongest vortex shedding, most prone to whistling | Span under 1200 mm, or switch to flat-oval, or insert one 30 mm bar every six |
| Square-tube grilles (e.g. 50 x 50) | Regular shedding at the corners, easily synchronised across neighbouring bars | Stagger the bars by about 20 mm front to back - very effective |
| Perforated aluminium screens | The hole array is itself a multi-slot structure and can produce edge tones | Use holes of at least 6 mm or at most 2 mm, avoiding the 3-5 mm sensitive band; infill any cavity behind with acoustic wool |
| Cable and rod grilles | Slender members vibrate readily and with large amplitude | Follow stay-cable practice with helical strakes or dampers; avoid dense parallel arrays |
5 common mistakes
- Treating the whistle as an assembly problem and re-torquing everything - torque does not change the shedding frequency. Measure first; it saves a full round of rework.
- Chasing perfectly equal spacing and equal diameter for appearance - this is the single biggest trigger. A 5% pitch variation is almost invisible and changes the behaviour by an order of magnitude.
- Using ordinary rubber as a damping pad - after two years it hardens, stops damping and becomes a new loose element. Use EPDM or neoprene outdoors, and make it replaceable.
- Pairing aluminium grilles with galvanised steel brackets and no insulation - contact corrosion shows within six months on coastal projects. Isolate the interface.
- Running static analysis only and skipping the wind-vibration check - the structural report covers strength and deflection but rarely asks whether the natural frequency sits in the shedding band, which is the only real criterion for whistling.
In one line: grille whistling is vortex-shedding frequency meeting natural frequency. Compute fs = 0.2V/D (a 50 mm bar at 10 m/s sheds at 40 Hz), then the natural frequency; if they fall within ±20% you get lock-in. At design stage the two most effective moves are breaking the equal-spacing rhythm and halving the span (which quadruples the frequency); on site work through measure - damping - strakes - supports - heavier bar. Putting this into the shop-drawing notes is far cheaper than returning to the scaffold afterwards.
Need a wind-vibration screen for your grille design?
Send us the bar section, spacing, span and project wind zone - we return a vortex-shedding frequency check, a lock-in risk assessment and the recommended geometry or damping detail for your shop drawings.
Foshan source factory · manufacturing since 1982 · Stainless · Liquid-cooling manifolds · Architectural metalwork
Contact Us