- November 19, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
The RECD has been fitted, the crew has packed up and the first loaded run shows a pressure reading that nobody expected. Perhaps the controller flags a high differential pressure within the first hour. Perhaps the engine’s own back-pressure alarm trips. Perhaps the set simply feels slow to take load. Whatever the symptom, the question is the same: where is the extra pressure coming from?
The reflex is to blame the new filter. Sometimes that is right. More often, the pressure is being lost somewhere along the exhaust path, and the filter is only the most visible suspect. A tight bend, a reducer, a retained silencer, a collapsed bellows, a closed bypass damper, a misrouted sensing tube or an instrument reading at the wrong load can each produce a number that looks like a blocked filter.
This guide gives you a method rather than a list of causes. First we separate the two pressures that people confuse. Then we build a pressure budget for the exhaust path, so that every component has a share of the total. Finally we walk through a nine-test ladder that finds the loss in the order that costs least. Throughout, the supplier’s commissioning data and the engine maker’s limits override any general number in this article.
Step zero: two different pressures
Almost every confusing conversation about back-pressure starts because two things are called by the same name.
Differential pressure (ΔP) across the filter is the difference between the pressure just before the filter and just after it. It measures what the filter itself is costing. Controllers and alarms usually report this. It is measured in millibar (mbar) or kilopascal (kPa).
Total exhaust back-pressure at the engine is the pressure the engine sees at its exhaust outlet, usually the turbocharger outlet or the manifold, relative to atmosphere. It includes the filter, the silencer, every pipe, bend and bellows, and the stack outlet. This is the number the engine maker limits.
A filter can have a perfectly acceptable ΔP while the engine still sees too much back-pressure, because the pipework or silencer is adding its own losses. The reverse is also possible: a high ΔP reading on a filter that is fine, caused by a gauge fault. You cannot diagnose anything until you know which of the two you are looking at.
DieselNet’s technology guide on diesel exhaust back pressure explains why an engine is sensitive to these losses: gas cannot flow against rising pressure, so the engine has to pump harder, and it publishes allowable limits that vary with engine size. Your engine’s own manual gives the limit that matters for warranty. Our article on how exhaust back-pressure affects a DG set describes the consequences of exceeding it, including loss of power, higher exhaust temperature and turbocharger stress.
The pressure budget: give every component a share
An engineer designing an exhaust system does not hope for the best. They add up the pressure drop of each element and check that the total is below the engine limit. When you are fault-finding, the same arithmetic is the quickest way to locate a problem. You compare each element’s expected drop with its measured drop.
The table below is an illustrative budget for a mid-sized set at rated load. The numbers are invented to show the method and are not data for any product. Your supplier’s drawing and the engine data give the real ones.
| Element | Expected drop (kPa, illustrative) | Measured drop (kPa) | Gap |
|---|---|---|---|
| Turbo outlet to first flange, including expansion piece | 0.4 | ||
| Flexible bellows | 0.2 | ||
| Pipe run to RECD, two bends | 0.8 | ||
| RECD (clean filter) | 3.0 | ||
| Pipe run after RECD to silencer | 0.5 | ||
| Silencer | 1.5 | ||
| Pipe to stack, one bend | 0.6 | ||
| Stack outlet / rain cap | 0.3 | ||
| Total at engine | 7.3 |
If the engine limit in this example were 10 kPa, the design would have about 2.7 kPa of margin. Now imagine the measured total is 12 kPa. The budget lets you ask the useful question: which line has a gap? A gauge placed at each flange gives you the drop for each section directly. If the RECD line reads 3.2 kPa and the silencer line reads 4.8 kPa, you have a silencer problem, not a filter problem, and cleaning or replacing the filter would have been wasted effort.
You do not need pressure taps everywhere to use this idea. Even three measurement points (engine outlet, RECD inlet and RECD outlet) split the system into three parts and tell you which part holds the excess.
Before you touch anything: freeze the evidence
A high reading is a clue, so capture it before changing anything.
- Record the operating state. Load in kW or percentage, engine speed, coolant and exhaust temperatures, ambient temperature and time since start.
- Record the readings. ΔP from the controller, total back-pressure at the engine (if a tap exists), and the time stamp of each. A handwritten note is fine.
- Photograph the installation. Pipe routes, supports, bellows, silencer, damper positions and the sensing tubes.
- Keep the commissioning data and drawing. Compare against them throughout.
- Note what changed. The fitting date, the parts replaced and any engine work done at the same time.
The reason is simple. Once someone tightens a flange or disconnects a tube, the original condition is gone, and you cannot tell which change mattered. Our checklist for the end of the job in closing an RECD punch list before handover lists the records worth capturing.
The nine-test ladder
Run the tests in order. Early tests are quick and cheap, and they can end the investigation. Each test has a question, a method, a result that implicates the suspect and a result that clears it.
Test 1: Is the gauge telling the truth?
Question. Does a second, independent instrument agree with the controller reading, at the same load?
Method. Compare the controller’s ΔP with a calibrated portable manometer or digital gauge connected at the same taps or at spare taps. Check that both are at stable load for at least a few minutes.
Implicates. A difference that cannot be explained by the instruments’ stated accuracy.
Clears. Agreement within tolerance.
Why first. It takes half an hour, and a wrong gauge sends every later test in the wrong direction. Our guide on separating a sensor fault from an RECD performance problem offers a fuller routine, and pressure sensors versus pressure switches explains the typical differences between instrument types.
Test 2: Are the sensing lines installed and routed correctly?
Question. Are the high-pressure and low-pressure tubes connected to the right taps, free of kinks, leaks and condensate, and routed away from heat?
Method. Trace each tube from tap to transmitter. A reversed pair can produce negative or nonsensical ΔP, while a leaking or blocked tube can produce a drifting or frozen value. Check for water in the line, particularly after a first run when new exhaust systems produce more condensation. The tube from the filter port to the transducer is a classic hidden fault in this kind of equipment, because a diagnostic tool will not reveal a crimped or leaking tube, a point made in a fleet-maintenance article on after-treatment faults.
Implicates. Reversed, kinked, leaking, wet or heat-damaged tubes.
Clears. Clean, correctly connected, dry tubes that respond when you apply a small test pressure.
Test 3: Is the load at the moment of reading the load you think it is?
Question. Is the reading being compared against a limit that applies to a different load?
Method. ΔP rises roughly with the square of exhaust flow, so a reading at 100% load can be several times a reading at 50%. Check the controller’s alarm limit and the supplier’s commissioning figure and make sure they refer to the same load band. See comparing RECD pressure readings at different generator loads.
Implicates. A mismatch between reading load and limit load.
Clears. A reading that exceeds the limit at the matching load.
Test 4: Is the filter itself in order?
Question. Is the new filter in the state it was shipped in?
Method. Compare ΔP at the commissioning reference load with the factory test or supplier datasheet. Ask whether the filter has been run at all: a fresh catalysed filter may release coating moisture or a first burn-off, but it should not carry a high ΔP for long. Check for obvious installation errors such as reversed flow direction (if the design is directional), packing material or protective caps left in, a damaged inlet face or an incorrectly seated canister. Our RECD delivery inspection checklist and the guide on packaging and transporting an RECD without damage describe what should be checked on arrival.
Implicates. A ΔP well above the factory figure at matching flow, or visible faults at the inlet.
Clears. A ΔP close to the factory figure at matching flow.
Test 5: Are there hidden restrictions in the pipework?
Question. Is something narrowing the path?
Method. Compare actual pipe diameters, reducers and bends with the general arrangement drawing. Look for pipe sizes reduced to fit an existing flange, mitred bends with sharp internal angles, a gasket protruding into the flow, a bellows that has collapsed or been fitted with its internal liner reversed, a partly closed damper, insulation or debris inside the pipe and any “temporary” blanking plate. Our article on RECD exhaust bends, reducers and layout review explains how layout affects loss, and flexible bellows in an RECD exhaust system covers the bellows.
Implicates. Any dimension that is smaller than drawn, any sharp or doubled bend, any visible obstruction.
Clears. A route that matches the drawing, with smooth bends and full-bore flanges.
Test 6: Is the old silencer still in the path, and in what condition?
Question. Did the retrofit leave an old silencer in series with the new device, and is it blocked, collapsed internally or simply too restrictive?
Method. Check the drawing and the site: some retrofits keep the existing silencer and add the RECD, and some replace it. If the silencer stays, find out its age and internal condition. Old silencers can contain collapsed baffles, carbon build-up or corroded internal parts. A pressure tap on each side of it gives its drop directly. Our guide on whether an RECD can replace a generator’s exhaust silencer discusses how the two interact, and reusing existing exhaust pipework in an RECD retrofit lists what to inspect in old pipework.
Implicates. A silencer drop well above its datasheet value, or visible internal damage.
Clears. A silencer drop close to its datasheet value.
Test 7: Is there a bypass or damper in the wrong position?
Question. If the system has a bypass, diverter or isolating damper, is it where it should be?
Method. Confirm damper position, actuator signal and limit switch status. A half-open diverter can force gas through a filter at a flow it was not meant to carry, or choke the main path. Operate the damper through its stroke under supervision and verify the indicator matches reality.
Implicates. A damper not reaching its commanded position, or a mismatch between indicator and actual blade.
Clears. Correct, repeatable damper position.
Test 8: Is the engine itself contributing?
Question. Is the engine producing more exhaust flow or soot than the design assumed?
Method. Compare the engine data you gave the supplier with the engine as it is. A changed turbocharger, an uprate, a different fuel map, a leaking injector or a failing air filter will change flow and soot. Run an opacity check and compare it with earlier records. Our guide on specifying an RECD from OEM exhaust flow data shows which engine facts matter, and the question of overhauling before fitting an RECD covers the timing.
Implicates. An engine whose flow or smoke is outside what the supplier assumed.
Clears. An engine consistent with the supplied data.
Test 9: Is the device the right size for the real duty?
Question. After everything above is clear, is the high reading simply the right answer for an undersized or mismatched device?
Method. Return to the supplier with a full data pack: engine data, drawing, measured pressure budget and load history. Ask them to compare the measured ΔP with their calculated figure for your flow and temperature, and to explain any difference. If the device is genuinely too restrictive, the options are a larger device, a change of filter core, a lower-resistance silencer or a relocated layout.
Implicates. A measured ΔP consistent with the supplier’s data but too high for the engine’s limit.
Clears. Not applicable: this is the last test, and a failure here points to a design change.
The ladder at a glance
| Test | Looks at | Typical time | What a hit tells you |
|---|---|---|---|
| 1 | Gauge accuracy | 30 to 60 min | Reading is wrong |
| 2 | Sensing lines | 30 to 60 min | Reading is wrong or drifting |
| 3 | Load matching | 15 min | Comparison is wrong |
| 4 | Filter condition | 1 to 2 hours | Filter, shipping or fitting fault |
| 5 | Pipework | 1 to 3 hours | Installation fault |
| 6 | Silencer | 1 to 3 hours | Old component problem |
| 7 | Bypass and dampers | 1 to 2 hours | Control or mechanical fault |
| 8 | Engine | Half day | Engine-side cause |
| 9 | Design sizing | Supplier review | Design change needed |
Notice how many of the nine tests have nothing to do with the filter. In practice, it is uncommon for a brand-new, correctly specified filter to be the cause of an immediate high reading.
What “high” means: setting the threshold before you start
Arguments about whether a number is high are common. Settle the threshold from documents, not opinion.
- Engine limit. The engine maker’s maximum allowable exhaust back-pressure at rated load, from the manual or data sheet.
- Supplier’s design figure. The calculated or tested ΔP of the clean device at your flow and temperature, and the alarm and action levels.
- Contract. Any back-pressure guarantee or acceptance criterion in the purchase order. See RECD performance guarantees and operating conditions and RECD trial installation acceptance criteria.
- Regulator or permit conditions. If any apply to your site, they define what you must demonstrate.
A US EPA technical bulletin on diesel particulate filter installation emphasises that a back-pressure monitoring system should be periodically inspected to confirm it is working, which is a reminder that the monitor itself is part of the system you are diagnosing.
Reading symptoms: what the pattern suggests
Different patterns of high pressure hint at different parts of the ladder. These are tendencies, not rules.
| Pattern | Likely region | Reason |
|---|---|---|
| High on the very first start, steady with load | Installation (tests 2 to 7) | A built-in restriction does not depend on soot |
| Low at light load, much too high near rated load | Pipework or silencer | Losses rise with the square of flow |
| Starts normal, climbs over the first hours | Filter loading, wet soot, moisture | Soot or condensate accumulating in a new system |
| Jumps in steps | Control or damper | Position changes |
| Wanders at constant load | Gauge or tubes | Instrument problem |
| High ΔP, normal engine back-pressure at the turbo | Filter or gauge | The loss is inside the RECD or in its reading |
| Normal ΔP, high engine back-pressure | Pipework, silencer or stack | The loss is outside the RECD |
| High in the morning, lower after warm-up | Condensate, moisture in a cold system | Water in the filter or sensing lines |
The last row deserves a note. A new installation, particularly in a damp plant room, can hold moisture in insulation, canning and ceramic. The first warm-up drives it out, and ΔP may be high at first and settle. If it settles after a loaded warm-up and stays settled, moisture was probably the cause. If it keeps high, move on. Our guide to protecting an RECD against rainwater and condensation explains how moisture gets in, and the cold-start article covers warm-up behaviour.
A worked example: the loss that was not in the filter
The following is an illustrative scenario, not a real customer. A factory fits an RECD to a 500 kVA set. On the first loaded run, the controller shows ΔP of 5.1 kPa at 80% load against a supplier figure of 3.4 kPa, and the engine’s own alarm reads high back-pressure. The site team assumes the filter is faulty.
Test 1. A portable gauge agrees within 0.1 kPa. The reading is real.
Test 2. Sensing lines are correct and dry.
Test 3. Both readings are at 80% load. The comparison is fair.
Test 4. The factory test sheet shows the filter at 3.3 kPa at equivalent flow. No obvious fitting faults.
Test 5. The drawing shows a 400 mm pipe throughout. On site, a 300 mm section was installed between the RECD outlet and the silencer inlet, because the silencer flange was smaller than drawn. A reducer and an expander sit either side of it. The team places taps across the section and reads a drop of about 1.6 kPa, against an expected 0.5.
Result. The “filter problem” was a pipe section. The pressure budget showed a gap on one line, and only that line. The contractor replaces the 300 mm section and its fittings with a full-bore transition to the silencer flange, which is made with a new adapter. On the next loaded run, ΔP across the RECD reads 3.5 kPa and total engine back-pressure falls below the limit.
Had the team replaced or cleaned the filter first, they would have spent days and money and found no change. The method, not the cause, is the lesson: measure the drop across each element and compare it with the budget.
What to ask the contractor and supplier
When a high reading shows up after installation, responsibility is often shared between the supplier, the installer and the customer. Having the facts in order avoids a dispute. Ask:
- Which drawing revision was built, and where does the as-built differ?
- What were the pipe sizes, bends and fittings actually installed?
- What was the factory ΔP of the filter at the test flow, and under what conditions?
- What was the measured ΔP and total back-pressure at commissioning, at what loads?
- Who supplied and fitted the silencer, bellows and dampers?
- Which pressure taps exist, and can additional taps be added?
- What does the contract say about back-pressure limits and remedies?
See customer and installer responsibilities in an RECD project for how responsibilities are normally divided, and RECD commissioning with load bank testing for how to capture a clean set of readings.
Safety notes
Exhaust systems are hot, and a running engine produces carbon monoxide. Never work on a running exhaust system without lock-out, and let the system cool before opening flanges. Do not disconnect pressure tubes on a running system unless the supplier’s procedure allows it, since hot gas and soot can escape. Wear appropriate protection when handling soot or insulation, and follow site permits for work at height. Our guide to safe access to RECD sampling ports on a tall chimney covers access.
Frequently asked questions
Is some increase in back-pressure normal after fitting an RECD?
Yes. A filter adds resistance to the exhaust path, so total back-pressure rises. What matters is whether the new total stays within the engine maker’s limit and whether the filter’s ΔP matches the supplier’s figure for your flow.
How can I tell whether the engine limit has been exceeded?
You need a measurement of absolute back-pressure at the engine’s exhaust outlet or turbo outlet, taken at rated load and compared with the engine data sheet. A controller’s ΔP alone cannot show it.
Can a new filter be blocked on day one?
It is possible, for example through packing material left in the inlet, a damaged inlet face or contamination during storage, but rare. A high reading on day one more often comes from the gauge, the tubes or the pipework. See storing an RECD before installation.
Should I run the set at full load to burn the high reading off?
Not without the supplier’s advice. If the cause is a restriction or a gauge fault, extra load will not help and may stress the engine. If the cause is moisture or first-run conditioning, the supplier can advise on a controlled warm-up.
Does a high back-pressure reading void the engine warranty?
It can complicate it, because engine makers set limits and may ask for evidence. Keep your measurements and the supplier’s data together. See will installing an RECD affect a generator’s OEM warranty.
Bringing it together
High back-pressure after an RECD installation is a location problem before it is a filter problem. Separate ΔP from total back-pressure, build a pressure budget so that each section has a share, and walk the nine-test ladder from the cheapest test to the most expensive. In many cases the loss turns out to be a gauge, a tube, a reducer or an old silencer, and only occasionally the filter.
If you would like help checking a pressure budget or comparing your readings with the figures for a particular engine, see our retrofit emission control devices, check the certifications or contact the Aceget team.
Related Reading
- How Exhaust Backpressure Affects a DG Set
- Common RECD Installation Mistakes to Avoid
- RECD Exhaust Bends, Reducers and Layout Review
- Comparing RECD Pressure Readings at Different Generator Loads
- RECD Commissioning with Load Bank Testing
[…] High Backpressure After RECD Installation: A Fault-Finding Guide […]