How Dilution Can Mislead RECD Before-and-After Measurements

Pour a glass of strong tea into a jug of water and it looks weaker. The amount of tea hasn’t changed, only how much water it’s mixed with. Diesel exhaust works the same way. Add clean air to it, and every concentration you measure (milligrams of particulate per cubic metre, parts per million of CO) drops, even though the engine is emitting exactly as much as before.

That’s why dilution matters so much in Retrofit Emission Control Device testing. Most RECD comparisons start with concentrations, and a concentration falls whenever extra air is added, whether or not the device is doing its job. If more air gets into the “after” sample than the “before” sample, the device gets credit for a reduction it didn’t make. If the extra air goes the other way, or carries its own pollution, a good device can be penalised.

This article covers six ways dilution misleads RECD before-and-after measurements, with a worked example for most of them and practical checks to prevent each.

Concentration vs Amount: The Core Problem

An RECD reduces the amount of particulate leaving the engine, measured as grams per hour or grams per kilowatt-hour. A sampling instrument measures concentration, in mg/m³. The two are linked by exhaust flow:

Amount (g/h) = Concentration (mg/m³) × Flow (m³/h) ÷ 1,000

If air gets into the exhaust, concentration goes down and flow goes up by the same proportion, so the amount stays the same. Trouble starts when someone compares concentrations without accounting for how much air each one contains, or multiplies a diluted concentration by an undiluted flow figure.

Every dilution problem below is a version of that mistake. For a fuller explanation of units, see RECD emissions data in mg/Nm³ vs g/kWh.

Where the Extra Air Comes From

Not all dilution is accidental, and not all of it is a problem. It helps to map where air can enter between the engine and the measurement:

Where air enters	Intended?	Effect if not accounted for
Combustion (excess air in the engine)	Yes, normal diesel operation	Varies with load; see load matching
Leaks upstream of the RECD	No	Dilutes both upstream and downstream samples; minor bias
Leaks between the RECD and the downstream port	No	Dilutes only the “after” sample; flatters the device
Sampling dilution tunnel	Yes, deliberately	Only accurate if the dilution ratio is known exactly
Dilution air carrying its own PM	No	Adds particulate; penalises the device at low readings
Ambient air mixing with the plume outside the stack	Inevitable	Makes plume readings useless for RECD verification

Trap 1: Air Leaks Between the RECD and the Sampling Port

How it happens

A loose flange, a cracked flexible bellows, an open drain point or a poorly sealed port downstream of the device can all let air into the exhaust. The engine’s exhaust system often runs slightly below atmospheric pressure at certain points, especially near the stack top or where the gas has cooled, so air is drawn in.

Because the leak is after the RECD, only the post-RECD sample is diluted. The baseline isn’t.

Worked example: correcting with CO₂

A test at matched 60 kW load gives:

	PM (mg/m³, dry)	CO₂ (%, dry)
Before RECD	90	9.0
After RECD	24	6.0

Uncorrected reduction: (90 − 24) ÷ 90 = 73.3%, an apparent pass.

The CO₂ readings are the giveaway. At the same load, burning the same fuel, the engine produces the same CO₂ concentration in its exhaust. An RECD doesn’t absorb CO₂, and its oxidation catalyst adds only a negligible amount by converting CO and HC. A fall from 9.0% to 6.0% means the downstream sample is one-third ambient air.

Correct the after reading to the same CO₂ basis as the before reading:

After (corrected) = 24 × (9.0 ÷ 6.0) = 36 mg/m³

Corrected reduction: (90 − 36) ÷ 90 = 60.0%, a clear fail.

The leak inflated the result by more than 13 percentage points. Without the CO₂ measurement, nobody would have known.

How to prevent it

• Inspect and leak-check all joints between the RECD and the downstream port before testing. See flexible bellows in an RECD exhaust system
• Measure O₂ or CO₂ at the same port and the same time as each PM sample, in both tests
• Treat any unexplained difference in CO₂ or O₂ between before and after at the same load as a red flag, and find the cause before accepting the result

The instrumental O₂/CO₂ approach used in stack testing is described in US EPA Method 3A. Indian stack-monitoring labs use equivalent practice.

Trap 2: A Dilution Tunnel Ratio That Isn’t What It Says

How it happens

Many labs pass the exhaust sample through a partial-flow dilution system before it reaches the filter. A small, measured stream of exhaust is mixed with a larger, measured stream of clean, filtered air. This cools the sample, controls condensation and mimics what happens as exhaust meets the atmosphere. The raw concentration is calculated by multiplying the measured (diluted) value by the dilution ratio (DR).

Any error in DR passes straight through to the result. If the DR is 10% higher than assumed, the raw concentration is understated by 10%. Getting the dilution ratio right within tight limits has long been recognised as one of the critical requirements for valid partial-flow particulate measurement under ISO 8178. See, for example, SAE Technical Paper 932466.

DR can drift for several reasons:

• Flow controllers calibrated for air but measuring exhaust gas, which has a different composition
• Changes in exhaust pressure at the probe. An RECD adds back-pressure, which can change how much exhaust the probe draws, so the DR may differ between the before and after tests even at the same setpoint
• Partly blocked sample lines or probes
• Temperature changes affecting flow measurement

Worked example: a tracer check catches the drift

A lab runs the tunnel at a nominal DR of 15 for both tests.

• Before: filter reads 4.5 mg/m³ diluted × 15 = 67.5 mg/m³ raw
• After: filter reads 1.2 mg/m³ diluted × 15 = 18.0 mg/m³ raw
• Reported reduction: (67.5 − 18.0) ÷ 67.5 = 73.3%

Now apply a CO₂ tracer check during the after test. Ambient CO₂ is 0.04%, raw exhaust CO₂ is 7.54%, and CO₂ in the diluted sample is 0.40%.

DR (actual) = (7.54 − 0.04) ÷ (0.40 − 0.04) = 7.50 ÷ 0.36 = 20.8

The tunnel was diluting much more than its setpoint suggested. The correct raw concentration is:

After (corrected) = 1.2 × 20.8 = 25.0 mg/m³

Corrected reduction: (67.5 − 25.0) ÷ 67.5 = 63.0%, a fail.

The filter weights were right. The multiplier was wrong.

How to prevent it

• Require a measured dilution ratio, not just a setpoint, for every sample in both tests
• Use a tracer gas check (CO₂ or NOx in raw and diluted streams) as an independent confirmation of the flow-based DR
• Keep the same DR setpoint for the before and after tests, and report the measured DR for each
• Ask the lab how DR is controlled when exhaust back-pressure changes

Trap 3: Dirty Dilution Air

How it happens

The air used for dilution, whether in a tunnel or leaking into the stack, isn’t particle-free unless it is properly filtered. In Delhi NCR during winter, ambient PM2.5 can sit at several hundred micrograms per cubic metre for days. See why Delhi AQI increases in winter and PM2.5 vs PM10.

In the baseline test, the exhaust carries so much particulate that the dilution air’s contribution is negligible. After a good RECD, the exhaust is much cleaner, so the same background becomes a meaningful share of what lands on the filter.

Worked example: background on a polluted day

Suppose the dilution air hasn’t been HEPA-filtered and carries 0.10 mg/m³ of PM. The tunnel runs at a measured DR of 20, so about 95% of the gas passing the filter is dilution air. That adds roughly 0.095 mg/m³ to every diluted reading.

	Diluted reading (mg/m³)	Background share	Raw result if background ignored	Raw result after subtracting background
Before	4.00	~2%	80.0	78.1
After	1.10	~9%	22.0	20.1

• Reduction ignoring background: (80.0 − 22.0) ÷ 80.0 = 72.5%
• Reduction after subtracting background: (78.1 − 20.1) ÷ 78.1 = 74.3%

This time dilution penalised the device, by nearly two points. On dirtier days, with higher DR or a very efficient RECD, the effect is bigger.

How to prevent it

• Use HEPA-filtered dilution air in any dilution system
• Collect a dilution-air background sample and subtract it, as the lab’s procedure specifies
• Take field blanks (filters handled in exactly the same way but not exposed to exhaust) to capture weighing and handling errors
• When the post-RECD filter mass is very small, sample for longer so the measured mass is well above the background and the balance’s limit

Trap 4: Correction Factors That Amplify Error at Light Load

How it happens

To put concentrations on a common basis, labs often correct them to a reference oxygen level:

C_ref = C_measured × (20.9 − O₂_ref) ÷ (20.9 − O₂_measured)

This works well when exhaust O₂ is moderate. At light load, diesel exhaust can contain 16–18% O₂ or more, because the engine burns little fuel in a full charge of air. The denominator (20.9 − O₂) becomes small, and small errors in the O₂ reading turn into large errors in the corrected result.

Worked example: half a percent of O₂

At a light-load test point, correcting to 15% O₂:

• If the analyser reads 18.0% O₂: factor = 5.9 ÷ 2.9 = 2.03
• If the true value is 18.5% O₂: factor = 5.9 ÷ 2.4 = 2.46

A 0.5 percentage-point O₂ error, well within what a poorly calibrated or drifting analyser can produce, changes the corrected concentration by about 21%. If that error affects only one of the two tests, the reduction figure moves by a similar amount.

At a mid-load point with 11% O₂, the same 0.5-point error changes the factor by only about 5%.

How to prevent it

• Calibrate O₂ and CO₂ analysers with certified gases on the test day, and record the checks
• At light-load points, consider CO₂-based normalisation, or compare on a mass-rate basis (g/h or g/kWh) using flow measured at the same port
• Report the raw O₂/CO₂ values alongside the corrected results so the correction can be checked
• Treat light-load corrected results with extra care, and run repeats

Light-load test points matter for many standby sets. See choosing an RECD for a generator running at low load.

Trap 5: Mixing Wet and Dry Bases

How it happens

Diesel exhaust contains a significant amount of water vapour from combustion, often several percent by volume. Concentrations and flows can be expressed on a wet basis (including the water) or a dry basis (with it removed). Neither is wrong, but mixing them is.

Removing water is a kind of “reverse dilution”: the same particulate in a smaller dry volume gives a higher concentration. If one test reports dry concentration and the other reports wet, or if a dry concentration is multiplied by a wet flow, the error can easily be 5–10%.

How to prevent it

• State the basis (wet or dry) for every concentration and flow in the report
• Make sure the before and after tests use the same basis
• Make sure any conversion to g/h or g/kWh uses a concentration and a flow on the same basis and at the same reference temperature and pressure (check whether “normal” means 0 °C or 25 °C)

Our guide on how to calculate RECD particulate reduction walks through the conversion arithmetic step by step.

Trap 6: Dilution That Changes the Particles Themselves

How it happens

Dilution doesn’t just make particles less concentrated. It can change how many particles exist and how much they weigh. As hot exhaust is cooled and diluted, some vapours (organic compounds from fuel and oil, and sulfuric acid) condense onto existing particles or form new ones. How much of this happens depends on:

• The dilution ratio
• The temperature at which dilution happens
• How long the diluted sample takes to reach the filter
• How much “soot surface” is available for vapours to condense on

After an RECD removes most of the soot, there’s much less surface for vapours to condense on, so condensation behaviour downstream of the device can differ from upstream. Changing dilution conditions between tests changes how much of this condensable material ends up on the filter.

This effect matters more for particle number than for particle mass, but it can still shift mass results noticeably at low post-RECD levels, especially where lube oil consumption is high. See engine oil consumption and RECD suitability.

How to prevent it

• Keep dilution ratio, dilution temperature and residence time the same in both tests
• Use the same sampling equipment and settings for upstream and downstream samples
• Note any visible condensation or wet filters, and ask the lab to comment on them

Sampling location also affects this, because gas temperature at the port controls how much material has already condensed. See how exhaust sampling location affects RECD test results.

Why Plume Readings Can’t Be Corrected

Some people try to verify an RECD by measuring the plume outside the stack with a portable particle monitor. Here, dilution is uncontrolled, changes with every gust of wind, and can’t be measured. The exhaust might be diluted 50 times at one moment and 500 times the next, and ambient background is mixed in throughout. No correction can rescue those numbers. We explain why in can a handheld AQI monitor verify RECD performance.

What a Dilution-Proof Report Should Contain

Use this list when commissioning a test or reviewing a report:

Item	Why it’s needed
O₂ and CO₂ at each PM sampling point, with time stamps	Detects air ingress; enables correction
Analyser calibration records (zero and span gases)	Confirms the O₂/CO₂ values can be trusted
Measured dilution ratio for each sample (if a tunnel is used)	Shows the multiplier is real, not a setpoint
Tracer gas DR check	Independent confirmation of the dilution ratio
Dilution-air background and field blank results	Separates background from engine emissions
Filter mass gains vs balance detection limit	Shows the after result is above the noise
Wet/dry basis and reference conditions for all values	Prevents basis mix-ups
Exhaust flow measured at the sampling port	Allows a mass-rate (g/h) result that dilution can’t distort
Leak-check record for the exhaust between RECD and port	Rules out downstream air ingress
Reference O₂ used for corrections, and the formula	Lets a reviewer recalculate

A report with all of these can be audited by anyone, from the buyer’s engineer to a Pollution Control Board officer. A report with none of them gives a single number with no way to check it. If you’re specifying a test, write these items into the purchase order. See RECD tender specifications with measurable performance requirements.

Field Checks Before Any Sample Is Taken

1. Walk the exhaust from the RECD to the downstream port. Look and listen for leaks at flanges, bellows, drains and port caps.
2. Compare upstream and downstream CO₂ (or O₂) at steady load with a quick analyser check. At the same load they should match closely. A clear difference points to a leak.
3. Check the dilution system: confirm the dilution air filter is in place and the flow controllers are calibrated.
4. Run a tracer DR check at the first load point, and repeat it if conditions change.
5. Prepare field blanks and plan background sampling.
6. Agree the basis (wet/dry, reference O₂, reference temperature) in writing before results are calculated.

Quick-Fire Myths About Dilution

“More dilution makes the measurement safer.” It protects equipment and controls condensation, but every extra unit of dilution makes the result more sensitive to DR errors and background.

“O₂ correction fixes any dilution problem.” Only if O₂ is measured accurately at the same point and time as the PM sample. At light load, small O₂ errors become large result errors.

“If the stack looks clean, dilution doesn’t matter.” Visual clarity says nothing about how much air has been mixed in. A well-diluted plume from a poor device can look perfectly clear.

“The supplier’s report is in g/kWh, so dilution can’t affect it.” g/kWh is immune to dilution only if the concentration and the flow were measured on the same basis at the same point. If the flow came from an OEM datasheet and the concentration from a diluted downstream sample, the error is still there. See using OEM exhaust flow data when specifying an RECD.

Dilution and the Other Test Variables

Dilution rarely acts alone. It interacts with every other variable in an RECD comparison:

• Why generator load must be matched in RECD comparison tests: load sets the baseline excess air
• How to account for warm-up time when testing an RECD: cold pipework and joints can leak more until they expand
• RECD testing during regeneration: what should be recorded: pressure changes during regeneration can alter leak rates and probe flows
• How exhaust sampling location affects RECD test results: what lies between the device and the port decides whether air can get in

For the overall framework, start with planning an RECD emissions test before and after installation.

Frequently Asked Questions

Does dilution always make an RECD look better?

No. Air leaking in after the device flatters it, because the “after” sample is diluted. Dirty dilution air penalises it, because background particulate inflates the cleaner “after” reading more. Dilution ratio errors can go either way. The common thread is that they make the result unreliable in an unknown direction unless they’re measured and corrected.

Is O₂ or CO₂ better for dilution correction?

Both work when measured accurately. CO₂ is often more robust at light load, where O₂ is high and the O₂ correction formula becomes very sensitive to small errors. Many labs measure both and use them as a cross-check.

Why should CO₂ be the same before and after the RECD?

At the same load and fuel, the engine produces the same CO₂. An RECD doesn’t remove CO₂, and its catalyst adds only a negligible amount by oxidising CO and HC. A clear drop in CO₂ downstream of the device therefore usually means air has been added.

Can I avoid dilution problems by testing raw exhaust without a tunnel?

Raw-exhaust sampling removes the tunnel DR as a source of error, but leaks, wet/dry basis and condensation still apply. It also has its own challenges with high temperatures and condensation in the sampling train. The method should follow the lab’s accredited procedure, and the same method must be used for both tests.

Who should I ask to check for dilution problems?

Ask the testing lab, ideally one accredited by NABL for stack emission monitoring, to include the items in the report table above. Your RECD supplier should help verify leak-tightness of the installation.

Keep the Air Out of Your Result

A dilution error doesn’t show up as an obvious mistake. It shows up as a believable number that is wrong. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA. We build leak-tight installations and help clients plan O₂/CO₂ measurement, port locations and reporting requirements before acceptance testing. For the regulatory framework, see the CPCB website and our DG set emission regulations page. When you’re ready, contact our team.