- November 19, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
A whistle is a strange kind of fault. It rarely stops the set, it often comes and goes, and it is easy to describe wrongly. One person hears a leak, another hears a turbo, a third insists it is “the new filter singing”. Yet a whistle is also one of the most informative sounds an exhaust system makes, because the physics behind it is simple: a whistle is made when a flow of gas meets a small opening, a sharp edge or a resonant cavity. Find the opening, edge or cavity and you have found the fault.
This article treats the sound itself as the instrument. You will learn what makes an exhaust whistle, how to record the whistle so that its behaviour does the diagnosis for you, and how to separate the three main families of cause: a leak, a flow restriction and a loose or resonant part. We then walk through a listening routine, a lookup table that links pitch behaviour to likely causes, and the repairs and cautions that follow. Where a point depends on your device, the supplier’s guidance comes first.
What makes an exhaust whistle at all
Exhaust gas at a retrofit installation is hot, fast and pulsing. A tone appears when that flow interacts with geometry in one of four ways.
Jet tone. Gas escaping through a small hole or crack forms a thin, fast jet. The jet’s turbulence at the edge of the hole produces a hiss, and when the hole is narrow and clean-edged, a tone. The pitch generally rises as the pressure behind the hole rises, which is why a leak whistle often climbs with load.
Edge tone. Gas flowing past a sharp edge, such as a gasket that protrudes into the pipe, a lip on a reducer or the edge of a perforated plate, sheds small vortices at a regular rate. If that rate matches a resonance in the pipe or a plate, a clear tone results. The pitch tends to rise with flow velocity.
Cavity or pipe resonance. A closed side branch, a sensor port, a drain stub or a section of pipe between two obstructions can resonate like an organ pipe or a bottle. These tones are at fixed frequencies set by dimensions, and they appear only when the flow excites them, often in a narrow band of load.
Flutter. A thin loose component, such as a damper blade, a flap, a thin heat shield or a perforated liner edge, can vibrate in the flow. The sound may be a whistle, buzz or chatter, and it often depends on the angle of the part to the flow.
Not all whistles come from the exhaust. A turbocharger can whistle, particularly as boost rises, and a leak in the charge-air path can whistle on its own. A belt can squeal. A loose cable gland can sing in a draught. Sorting the source comes first, and the routine below does that.
Whistle, hiss, hum, rumble: naming the sound
Descriptions matter, so settle on a vocabulary before you start.
| Word | What it usually sounds like | Often associated with |
|---|---|---|
| Whistle | A steady, high, narrow tone | Small leak, edge tone, a resonating stub |
| Hiss | Broad, noisy, like escaping air | A larger leak, turbulent jet |
| Whine | Rising and falling pitch with engine speed | A turbocharger, belt-driven part or generator fan |
| Hum or drone | Low, steady tone, felt as well as heard | Pipe resonance, mounting vibration |
| Chatter or buzz | Rapid, irregular tapping | A loose thin part, damper flutter |
| Puff or tick | Rhythmic, in time with the engine firing | A leak at a joint fed by pulses |
An audio recording on a phone, taken at a fixed distance, is useful, since you can replay it and use a free spectrum app to read the dominant frequency. You do not need accuracy to the hertz; knowing roughly whether the tone is at, say, 1 kHz or 4 kHz helps, as does seeing whether it moves.
The three questions that do most of the work
If you answer only these three, you will have narrowed the cause substantially.
Question 1: Does the pitch follow engine speed, or does it follow load?
On a generator set, engine speed is held nearly constant at rated frequency, so pitch that tracks load is the useful variable. As load rises, exhaust flow and pressure rise.
- A leak whistle typically rises in pitch and loudness with load, because higher pressure drives a faster jet.
- An edge tone also tends to rise with flow velocity, but it may jump between steps as the flow locks onto a different resonance.
- A cavity resonance tends to stay at one pitch, sometimes appearing only in a certain band of load and vanishing outside it.
- A turbo whine follows turbocharger speed, which follows boost and so also load, but it usually sounds more like a smooth siren than a pure whistle.
If the set is on a variable-speed arrangement or if you can test during run-up, note how the pitch moves as speed changes. A whistle that tracks the engine speed rather than load suggests a mechanical source, such as a rotating component.
Question 2: Does the sound depend on temperature?
Run the set from cold and note the sound at start, after ten minutes and when fully warm.
- Present when cold, gone when hot. Metal expands as it warms, so a gap at a cold joint can close. Many gasket and flange leaks are loudest at start-up for this reason. Cold sounds can also come from condensate in a small orifice.
- Absent when cold, present when hot. A hot-only whistle often means thermal growth has opened a gap, or an expanding part has created an edge. It also fits flutter of a heat shield that has distorted when warm.
- Present throughout. A steady source such as a fixed hole or a fixed resonance.
Because sound speed rises with temperature, a resonance tone also shifts upward as the gas heats, so a rising pitch during warm-up on a constant load is consistent with a resonance.
Question 3: Does it change when something changes in the system?
If the plant allows, run controlled changes and listen.
- A regeneration cycle starts. A heater-assisted cycle changes temperature without much change in flow. A tone that changes at that moment may be linked to the heater, the air path or a sensor.
- A damper moves. If the system has a diverter or bypass, a whistle that appears or stops when the damper moves points to that part.
- A sensor is removed or capped. A tone from a sensing port or a sampling port changes immediately when the port is capped. This should only be done with the supplier’s approval and on a cold system, then with the set run again.
- Load is held steady for several minutes. A fixed whistle at fixed load suggests a steady source. A wandering tone suggests flutter or unstable flow.
The sound log: a one-page method
A written sound log turns a vague complaint into evidence the supplier can use. Copy the table below into a spreadsheet or print it, and fill in one row per observation, with the same person listening from the same position each time.
| Field | What to write |
|---|---|
| Date, time, set ID | |
| Time since start (min) | |
| Load (kW or %) | |
| Ambient temperature and weather | |
| Listening position (distance, direction from the RECD) | |
| Sound word (whistle, hiss, hum, chatter, tick) | |
| Pitch (low, mid, high, or a number from an app) | |
| Loudness (1 faint to 5 loud) | |
| Does it change with load? | |
| Does it change with temperature? | |
| Any event at the time (regeneration, transfer, damper move) | |
| Recording file name |
Three to five rows at different loads and times are enough to see a pattern. Our article on what to record in an RECD event log shows how to keep such notes with the rest of the asset records.
Family 1: leaks
A small leak is the most common source of a new whistle after installation or service. A joint was opened, a gasket was reused or a fitting was refitted, and now a thin path exists.
What makes a leak whistle plausible.
- It appeared right after work on the exhaust or sensors.
- The pitch rises with load.
- It is loudest at or near one joint, and a hose to the ear tracks it down.
- Soot streaks or a clean scoured patch appear near the source. See our guide on exhaust leaks around an RECD for the station-by-station walk.
Typical locations. Pressure tap fittings, sensor compression fittings, sampling port caps, gasket edges at flanges, bellows convolutions, drain plugs, door gaskets and welds near supports.
Why sensor ports matter. A tiny leak at a pressure tap or sampling port is a classic whistler because the orifice is small and the pressure is high. It also changes your pressure reading, so a new whistle and a new drift in ΔP together are a strong hint. See separating a sensor fault from an RECD performance problem.
Family 2: flow restrictions and edges
Here, there is no hole to the outside. The sound comes from gas squeezing through a narrow gap or shedding vortices from an edge inside the pipe.
What makes a restriction whistle plausible.
- The tone appears in a particular band of load and may disappear above or below it.
- ΔP or total back-pressure is higher than expected, or the engine feels sluggish.
- Inspection shows a reducer, a gasket lip, a protruding sensor, a partly closed damper or a blocked perforated element.
- The pitch rises smoothly with load or jumps in steps.
Typical locations.
- A gasket that protrudes into the pipe at a flange. It forms a sharp inward edge.
- A damper or diverter that is not fully open and leaves a narrow slot.
- A sensor probe that extends too far into the stream.
- A perforated tube or plate in a silencer or in the RECD where some holes are blocked with soot, so the remaining holes see higher velocity.
- The inlet face of the filter or catalyst if it is partly blocked, causing high velocity through the open channels. Truck-maker bulletins describe noisy after-treatment units in which the inlet face of a catalyst has become partly plugged and the noise falls once it is cleaned. These are vehicle systems, but the principle carries over.
Compare the sound with the pressure budget. Our guide on high back-pressure after RECD installation explains how to measure the drop across each section. A restriction whistle plus an excess drop across one section tells you where to look.
Family 3: loose, thin and resonant parts
The third family is not a leak or a restriction. A part is vibrating in the flow.
What makes a flutter or resonance plausible.
- The tone is irregular or buzzy, or changes when you press on a heat shield or tap the body.
- It depends on temperature, since parts distort when hot.
- It is loudest at a particular load, where the pulse frequency of the engine matches a part’s natural frequency.
- There are loose heat shields, covers or brackets nearby.
Typical sources.
- A heat shield or cover that has come loose or distorted.
- A damper blade that is loose on its shaft.
- A thin liner inside a bellows, or a bellows with a loose braid.
- A cable or conduit tapping against a pipe.
- A short stub or capped branch that acts as a resonator.
If the sound is more rattle than whistle, move to our guide on rattling inside an RECD, since loose internals call for a different response.
Lookup table: pitch behaviour to likely cause
Use the table after your sound log has at least three rows. The right-hand column is a list of suspects to check, not a verdict.
| What the sound does | Most likely family | First checks |
|---|---|---|
| Whistle appeared after work on a joint or fitting; rises with load | Leak | Re-inspect the joints and fittings that were disturbed |
| Loud at start, fades as the system warms | Leak at a cold gap, or condensate | Check flanges and gaskets; re-torque when hot and cold as advised |
| Absent cold, present hot | Thermal growth leak, or distorted part | Check supports, expansion allowance, heat shields |
| Present only in a narrow band of load | Cavity resonance or edge tone | Look for side branches, ports, protruding gaskets and sensor probes |
| Steady pitch, independent of load | Resonant stub or mechanical source | Check stubs, branches; verify it is not a fan or alternator noise |
| Pitch jumps in steps as load rises | Edge tone locking onto resonances | Inspect for sharp edges in the flow; check gasket lips |
| Buzzy, irregular, changes if you tap or press | Loose or fluttering part | Check shields, damper blades, liners, cable ties |
| Rising smooth siren with boost | Turbocharger or charge air leak | Hand the check to the engine specialist |
| Appears with high ΔP | Restriction | Do the pressure budget and compare sections |
| Appears only when a regeneration heater runs | Heater air path, sensor or damper | Check heater housing, air path, seals |
A listening routine that does not need special tools
Safety comes first. Exhaust systems are hot and moving parts are close. Never put a hose, hand or tool near the exhaust outlet or a rotating part, and use hearing protection in the plant room.
- Choose a quiet time. Background noise hides small leaks, so pick a moment when other equipment is off.
- Warm up, then hold steady load. A stable load makes the sound easier to track. Use a load bank if needed.
- Listen from a fixed position first to hear the overall pattern, then move closer.
- Use a length of hose or a rolled sheet of card held to the ear as a simple directional device, with the open end pointed toward the suspected source at a safe distance.
- Use an ultrasonic leak detector if you have one. It filters out low-frequency plant noise and points at high-frequency jets.
- Compare the left and right sides of a joint, or compare with a similar joint you know is good.
- Record audio at each step and note the position.
- Stop, cool, inspect. Never poke at a hot joint. Mark suspected points with a paint pen when cold.
Do not use soapy water on a hot exhaust. If a cold, pressurised test is needed, take the supplier’s guidance on how to do it safely.
Ruling out sounds that are not the exhaust
Before opening a joint, check that the sound is really from the exhaust.
- Turbocharger. Listen at the turbo while the set runs, noting that the sound should follow boost. Ask an engine specialist to check the charge-air path for leaks. A hose clamp that has worked loose can produce a clear whistle.
- Air intake. A restricted air filter or a leaking intake joint can whistle. See generator air filter maintenance.
- Cooling fan and belt. A squeal at start or on load may be a belt. See how to prevent generator vibration issues for the related vibration checks.
- Alternator or enclosure. Canopy panels, louvres and cooling air paths can whistle in a strong draught.
- RECD cooling or heater blower, if fitted. A blower bearing or loose cover can produce a tone that is unrelated to exhaust flow.
If the sound does not change with the exhaust path, disconnect nothing; ask the engine specialist to help.
A worked illustration: three whistles, three answers
The following is an illustrative scenario, not a real site. After an RECD installation, three people in the facilities team report a whistle from the same 250 kVA set, each describing it differently.
Observation 1. The sound log shows the whistle appears at about 40% load, rises as load increases to 70% and disappears above that. It is absent at idle. At 70% load the listener with the hose finds the sound loudest near the pressure tap on the RECD inlet side.
Reading it. Rising with load and localised near a tap suggests a leak at a small orifice. Opening the plant room panel, the technician finds that the tap’s compression fitting is finger-tight after a calibration visit. Tightened to the supplier’s torque, the whistle stops, and the ΔP reading, which had been drifting low, returns to its earlier value.
Observation 2. A second whistle remains at a steady pitch at 30 to 40% load only. The sound log shows it stays at one pitch that does not rise with load. A spectrum app shows a narrow peak. A sample port stub on the stack, capped with a loose-fitting cap, is found. The stub is about 150 mm long, which suggests a cavity resonance. Fitting a properly seated cap and, with supplier agreement, shortening the stub removes the resonance.
Observation 3. The third sound is a buzz that appears only when hot. A heat shield on the silencer has distorted and is vibrating against its bracket. Re-fixing it with a rigid bracket and a spacer ends the buzz.
Three people, three words for the sound, three different causes. The log and the lookup table separated them in about two hours.
When a whistle is a warning to escalate
Most whistles are nuisances that can be fixed at a convenient time. Escalate quickly if:
- The whistle comes with a rise in ΔP or in engine back-pressure.
- There is soot staining around a joint that is growing.
- There is an exhaust smell in the plant room, or anyone feels unwell.
- The sound is accompanied by a visible gas jet or a very loud hiss.
- The whistle started suddenly after a regeneration, a load step or a mechanical impact.
- You suspect the whistle comes from inside the RECD and not from the pipework.
In those cases reduce load if safe, follow your site procedure and contact the supplier with the sound log, recordings and photographs. Our alarm response matrix provides a framework for deciding what to do.
Preventing new whistles
- Torque and re-torque fittings and flanges to the specified values, and record them.
- Replace gaskets rather than reusing them, and fit them so that they do not protrude into the flow.
- Cap and seat sampling ports properly after tests.
- Keep branches and stubs short, or have the supplier check them for resonance.
- Secure shields and covers with rigid fixings that tolerate heat.
- Check dampers for full travel and position.
- Take a baseline sound recording at commissioning, at a fixed position and load, so that a change is easy to hear later. This is an easy addition to RECD commissioning with load bank testing.
Frequently asked questions
Is a faint whistle from a new RECD normal?
Not usually. A new, well-installed device should be quiet in the sense that it adds no new tonal noise. A faint tone is worth tracking down because small leaks and edges tend to worsen with heat cycles.
Can the filter itself whistle?
A filter body is not a whistle, but a partly blocked inlet face or a damaged inlet can create high-velocity jets that make noise. The more usual sources are the fittings, gaskets, dampers and resonant stubs around it.
Why does the whistle change with the weather?
Temperature changes the speed of sound and the thermal growth of parts. Cold mornings may open a gap that closes when warm, and humid conditions can leave condensate that changes small orifices.
Does a whistle affect emissions or test results?
A leak near a sampling port or an upstream joint can change the sample and the pressure readings. If a test is planned, fix leaks first. See exhaust sampling location and RECD test results.
Should I tape or plug the suspected leak to test?
No. Tapes and plugs on hot exhaust parts fail, can be a hazard and destroy evidence. Mark the suspected point and inspect when cold.
Bringing it together
A whistle after an RECD installation is a clue with a shape: pitch, load, temperature and position. Answer three questions (what does the pitch follow, does temperature change it, and what changes when something in the system changes), keep a simple sound log, use the lookup table and sort the cause into leak, restriction or loose part. Fix the cause, record the torque and the date, and take a baseline recording so that you will know next time.
If you would like help interpreting a sound log or checking an installation for resonant stubs and edges, see our retrofit emission control devices, read the FAQ or contact the Aceget team.
Related Reading
- Exhaust Leaks Around an RECD: Symptoms and Inspection Points
- High Backpressure After RECD Installation: A Fault-Finding Guide
- Flexible Bellows in an RECD Exhaust System
- How to Prevent Generator Vibration Issues
- RECD Event Log: What to Record