RECD Emissions Test: How to Plan It, Measure It and Prove the Particulate Reduction

A Retrofit Emission Control Device is sold on one number: the share of particulate matter it removes. Under the CPCB framework, an RECD fitted to an in-use diesel generator has to show at least a 70% reduction in particulate matter without making other emissions worse. Much of the confusion comes after installation, when a facility wants to prove that number on its own DG set.

People run tests with the wrong instrument, at the wrong load, or report results in units that can’t be compared. Some take a phone photo of a clear chimney and treat it as proof. Others carry an air-quality monitor up to the stack. A few get a proper lab report and still can’t work out whether 68% or 72% is the true figure.

This guide covers the whole process in order. It starts with the planning decisions that make a before-and-after comparison valid. It then explains which measurement can actually judge an RECD, why consumer AQI monitors are the wrong tool, how to do the reduction calculation correctly, and why the unit on the report changes what the number means.

Quick answer: A credible RECD emissions test measures particulate mass, not just smoke opacity. It samples the exhaust before and after the device, or with and without it, at the same engine loads. It corrects for dilution and states the result in units that link emissions to engine work (g/kWh) or to a corrected concentration (mg/Nm³ at a reference O₂). Handheld AQI monitors can’t do any of this.

If you are new to how these devices trap soot, start with our explainer on what an RECD is and how it works.

---

Part 1: How to Plan a Before-and-After RECD Emissions Test

Most bad RECD data starts before anyone touches a probe. The comparison only means something if the “before” and “after” engines are, in every way that affects emissions, the same engine running under the same conditions. Planning is how you make sure of that.

Decide What Question the Test Must Answer

There are three different reasons to test, and each needs a different design:

1. Type verification: Does this RECD model meet the 70% requirement? This is done by CPCB-recognised labs such as ICAT or ARAI on a test engine over a defined cycle. You don’t repeat this at site. You check that the certificate covers your engine. Our guide on how to read an RECD test certificate shows what to look for.
2. Site acceptance: Is the device installed correctly and performing on my DG set? This is the before-and-after test most facilities need.
3. Periodic health check: Is the device still performing after months of use? This needs a baseline from the site acceptance test to compare against.

Write the purpose into the test plan. A site acceptance test that later becomes a dispute about guarantees is much easier to settle when everyone agreed in advance what was being measured. The difference between factory and site testing is covered in more detail in factory acceptance vs site acceptance tests for RECD projects.

Choose Your Comparison Method: Sequential, Bypass or Upstream/Downstream

There are three practical ways to get a “before” value:

Method	How it works	Strength	Weakness
Sequential (pre-install / post-install)	Test the DG set on bare exhaust, install the RECD, test again	Simple; matches how most projects run	Days or weeks pass between tests; engine condition, fuel and weather can change
Bypass	A valved bypass lets you route exhaust around or through the RECD on the same day	Same engine state, same fuel, same hour	Needs bypass hardware; the valve must seal fully
Simultaneous upstream/downstream	Two sampling ports, one before and one after the RECD, sampled at the same time or back-to-back	Best control of variables	Needs two properly located ports and possibly two sampling trains

Where the exhaust layout allows it, upstream/downstream sampling gives the cleanest result because the engine is identical for both readings. Port location matters a lot here. You need straight pipe runs for representative flow, and safe access for the sampling team. See our notes on safe access to RECD sampling ports on tall chimneys and on where to position an RECD along a long exhaust run.

Fix the Engine Condition Before the Baseline

If the engine is badly worn, the baseline is inflated. A high “before” number makes the reduction look better than it really is, and it can also overload the RECD. Before you test, check:

• The air filter is clean and intake restriction is within the OEM limit
• Injectors and the fuel pump have been serviced if smoke is visibly abnormal
• Lube oil consumption is normal (oil-derived particulate behaves differently in a filter; see engine oil consumption and RECD suitability)
• The exhaust has no leaks upstream of the sampling port, which would let in air and dilute the sample

If the engine is due for major work, do it first. Our guide on whether to complete an engine overhaul before fitting an RECD explains why.

Lock Down the Load Points

Emissions from a diesel engine change sharply with load. A test at 30% load before installation and 70% after it compares two different engines in practice.

For lab type-testing, the reference is the five-mode ISO 8178 D2 cycle for constant-speed gensets. It runs at 100%, 75%, 50%, 25% and 10% load, and each mode carries a weighting factor (DieselNet has a good summary of ISO 8178 cycles). Site tests often can’t reproduce every mode, because the building load may not allow it and a load bank may be needed. Whatever you choose:

• Use the same load points before and after, within a tight tolerance (±5% of rated power is a sensible target)
• Record actual kW, not just “about half load.” Use the DG controller or a power analyser, and log voltage, current and power factor
• Include the load where the DG set actually spends its life. For many standby sets, that is 30–50%

Your DG load logs are the best guide to which points matter. If the set mostly runs light, read choosing an RECD for a generator running at low load as well.

Control Fuel, Warm-Up and Stabilisation

• Fuel: Use fuel from the same batch for both tests if you can. Sulphur content affects sulphate particulate, so a change of supplier between tests can change the result. Background on this: how ultra-low sulphur diesel improves emission control.
• Warm-up: Run the set until coolant and exhaust temperatures are stable before sampling. Cold-start emissions are a separate question (see can an RECD work reliably during cold starts).
• Stabilisation at each mode: Hold each load until exhaust temperature stops drifting, typically several minutes, and sample only after that.
• New RECD conditioning: A brand-new filter substrate may behave differently in its first hours of operation. Agree with the supplier whether the “after” test happens immediately or after a short break-in run.

Record Back-Pressure Alongside Emissions

The second half of the CPCB condition, “without adversely affecting other emissions,” depends partly on how the engine breathes. Log exhaust back-pressure at each load point in both tests. A large rise is a warning about the installation, not just a number in a report. Read how exhaust back-pressure affects a DG set for the reasons.

A One-Page Test Plan Checklist

Before testing day, the plan should state:

• Purpose (site acceptance / periodic check)
• DG set make, model, kVA/kW rating, engine serial number, CPCB stage
• RECD model and serial number, certificate reference
• Comparison method (sequential / bypass / upstream-downstream)
• Sampling port locations, with straight-length details
• Load points in kW, tolerance, and how load will be applied
• Parameters: PM (gravimetric), CO, HC, NOx, O₂, CO₂, flow, temperature, back-pressure
• Sampling method and the lab’s accreditation
• Reporting units and the reference oxygen basis
• Number of repeat runs per load point
• Pass criterion and how it will be calculated

Writing this into the purchase order prevents arguments later. Our guide on RECD tender specifications with measurable performance requirements shows how to phrase it.

---

Part 2: Smoke Opacity vs Particulate Mass, Which Measurement Evaluates an RECD?

Both instruments look at the same exhaust, but they answer different questions. Using the wrong one is the most common technical error in field RECD verification.

What an Opacity Meter Actually Measures

An opacity meter shines a light through the exhaust and measures how much light gets through. The result is reported as percentage opacity or as a light absorption coefficient (k, in m⁻¹). It responds to anything that blocks light:

• Black carbon (soot) absorbs light very strongly, so it dominates the reading
• Water droplets and condensed fuel/oil vapour also scatter light (the “white smoke” and “blue smoke” situations; see our guides on white smoke and blue smoke)
• Very fine particles, below roughly 0.1 µm, block very little light per unit mass

Opacity is quick, cheap and good at catching a badly smoking engine. For engine diagnosis it is excellent. Black smoke from a generator shows up clearly on an opacity meter.

What a Gravimetric Particulate Measurement Measures

Gravimetric PM sampling pulls a measured volume of exhaust through a pre-weighed filter. The sample is drawn isokinetically in stack testing, which means at the same velocity as the gas in the pipe, so particles aren’t over- or under-collected. The filter is conditioned and weighed again. The mass gain divided by the sampled volume gives concentration. The method follows the same principles as US EPA Method 5 and the equivalent Indian stack-sampling practice. In lab dilution-tunnel testing, the filter captures the condensable fraction as well.

The result is a mass, and mass is what the regulation counts.

Why Opacity Fails as an RECD Yardstick

There are four specific reasons:

1. The floor effect. A well-functioning wall-flow filter type RECD often brings opacity down to near zero. If “before” is 8% and “after” reads 0.3%, the calculated reduction (96%) is dominated by meter resolution and zero drift. A reading of 0.6% would give 92%. The number tells you the exhaust is visibly clean, but it doesn’t tell you the mass reduction.
2. The particle-size blind spot. RECDs that use oxidation catalysts reduce the soluble organic fraction. That fraction adds a lot of mass but little opacity. Mass can therefore fall more, or less, than opacity suggests.
3. No standard conversion. Published correlations between opacity and PM mass exist, but they depend on the engine, the load and the fuel. Applying a generic factor to an RECD result adds an unknown error that is often larger than the margin you are trying to prove.
4. It isn’t the regulatory metric. The CPCB condition is stated in terms of particulate matter reduction. A report built only on opacity answers a question the regulator didn’t ask.

We discuss the wider point in does a smokeless DG set still need particulate measurement? The short answer is yes.

Side-by-Side Comparison

Factor	Smoke opacity	Particulate mass (gravimetric)
Output	% opacity or k (m⁻¹)	mg/Nm³, then g/h or g/kWh
Measures	Light extinction	Actual mass of particles
Time per reading	Seconds	30–60+ minutes per load point, plus lab weighing
Cost	Low	Higher (accredited lab, equipment, analysis)
Sensitive to fine/ultrafine PM	Weak	Captures mass of all sizes collected on filter
Useful after a high-efficiency RECD	Shows “clean,” cannot quantify %	Quantifies reduction
Accepted for 70% claim	No	Yes
Best use	Daily health check, troubleshooting	Acceptance testing, compliance evidence

The Practical Answer: Use Both, for Different Jobs

PM mass is the evidence. Opacity is the early warning. Take a gravimetric measurement at commissioning to establish performance. Then log opacity periodically as a cheap check that nothing has changed. A sudden rise in downstream opacity can mean a cracked substrate, a failed gasket or a bypass leak. Any of those calls for a proper inspection, and our RECD maintenance guide explains what to check.

---

Part 3: Can a Handheld AQI Monitor Verify RECD Performance?

No. It’s worth explaining why, because these monitors are cheap, widely available and show numbers that look precise.

How a Consumer AQI Monitor Works

Most handheld PM2.5 monitors use a small fan and a laser. Particles in the air scatter the light, and the monitor estimates particle count and size, then converts that to µg/m³ using a calibration made for typical outdoor or indoor air. The device then converts the concentration to an AQI category. If you want to know how that conversion works in India, see what AQI is and how it is calculated and PM2.5 vs PM10.

Five Reasons It Can’t Judge an RECD

1. The concentration scale is off by orders of magnitude.
Raw diesel exhaust particulate is typically tens of milligrams per cubic metre. That is tens of thousands of micrograms. Many low-cost sensors are specified for ambient levels up to around 1,000 µg/m³ and saturate well below exhaust levels. Even after a good RECD, undiluted exhaust can still be above the sensor’s linear range.

2. Hot, wet exhaust breaks the sensor and the reading.
Exhaust leaves the DG at a few hundred °C and is full of water vapour. As it cools at the sensor inlet, water condenses into droplets that the laser counts as particles. The heat can also damage the device. Neither effect produces a meaningful number.

3. The calibration is for the wrong aerosol.
Optical sensors assume a particular particle density, shape and refractive index. Fresh diesel soot is fractal, black and very fine, which is very different from the dust and secondary aerosol the sensor was tuned for. The mass estimate can be wrong by a large, unknown factor.

4. There is no flow measurement.
An RECD claim is about how much particulate the engine emits. That needs exhaust flow and engine power, and an AQI monitor knows neither. Holding it near the stack outlet measures a plume that is mixing unevenly with ambient air, which changes with every gust.

5. The ambient background confuses everything.
In Delhi NCR on a winter day, the ambient PM2.5 near your DG may already be several hundred µg/m³ (see why Delhi AQI increases in winter). A plume reading taken near the outlet can’t separate your engine’s contribution from the city’s.

What an AQI Monitor Can Reasonably Do

Used honestly, it has some secondary uses:

• Plant-room air quality: Checking whether exhaust leaks are raising PM levels inside a DG room or a basement installation
• Neighbour-complaint screening: Logging ambient levels at a boundary or near fresh-air intakes with the DG on and off, as a rough indication only
• Awareness: Showing occupants how ambient PM changes over the day

If you use one, read how to read an AQI app correctly to understand what the numbers do and don’t mean. None of these uses verifies the RECD. Only stack sampling of the exhaust itself, by an accredited lab, does that. A lab accredited by NABL for stack emission monitoring is the standard to look for.

---

Part 4: How to Calculate RECD Particulate Reduction from Test Results

The basic formula is simple. Most errors come from feeding it numbers that can’t be compared.

The Core Formula

PM Reduction (%) = [(PM_before − PM_after) ÷ PM_before] × 100

“PM” must be on the same basis for both readings, which means the same unit, the same correction and the same load. The three worked examples below show how the result changes depending on whether that rule is followed.

Worked Example 1: Raw Concentrations Can Mislead

A site test reports these numbers at nominal 50% load:

	PM measured (mg/Nm³, dry)	O₂ measured (%)
Before RECD	125	11.0
After RECD	35	12.5

Uncorrected reduction: (125 − 35) ÷ 125 = 72.0%. That looks like a pass.

The oxygen readings differ, though. A higher O₂ in the “after” test means the exhaust carried more excess air, either because the engine was at a lighter load than intended or because air leaked in around the RECD flanges. More air dilutes the particulate, which makes the concentration look lower than the engine’s true output.

Correct both to a common reference O₂. We use 15% here, which is common for engine exhaust, but any agreed reference works if both readings use it:

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

• Before: 125 × (20.9 − 15) ÷ (20.9 − 11.0) = 125 × 5.9 ÷ 9.9 = 74.5 mg/Nm³
• After: 35 × (20.9 − 15) ÷ (20.9 − 12.5) = 35 × 5.9 ÷ 8.4 = 24.6 mg/Nm³

Corrected reduction: (74.5 − 24.6) ÷ 74.5 = 67.0%. That falls short.

The device was the same and the samples were the same, yet the conclusion reversed. This is why an RECD test report should always state the O₂ (or CO₂) readings and the reference basis. If it doesn’t, ask for them before accepting the result.

Worked Example 2: The Weighted Multi-Mode Calculation (g/kWh)

For a full evaluation across the load range, the ISO 8178 D2 approach weights each mode. Take a 125 kVA (100 kW) DG set:

Mode	Load	Power (kW)	Weight	PM before (g/h)	PM after (g/h)
1	100%	100	0.05	30.0	7.5
2	75%	75	0.25	21.0	5.4
3	50%	50	0.30	16.0	4.3
4	25%	25	0.30	12.0	3.6
5	10%	10	0.10	9.0	3.1

Weighted specific emission = Σ(PM_i × W_i) ÷ Σ(P_i × W_i)

Denominator (same for both):
(100×0.05) + (75×0.25) + (50×0.30) + (25×0.30) + (10×0.10) = 5 + 18.75 + 15 + 7.5 + 1 = 47.25 kW

Before numerator:
(30×0.05) + (21×0.25) + (16×0.30) + (12×0.30) + (9×0.10) = 1.5 + 5.25 + 4.8 + 3.6 + 0.9 = 16.05 g/h
→ 16.05 ÷ 47.25 = 0.340 g/kWh

After numerator:
(7.5×0.05) + (5.4×0.25) + (4.3×0.30) + (3.6×0.30) + (3.1×0.10) = 0.375 + 1.35 + 1.29 + 1.08 + 0.31 = 4.405 g/h
→ 4.405 ÷ 47.25 = 0.093 g/kWh

Weighted reduction: (0.340 − 0.093) ÷ 0.340 = 72.6%

Look at the individual modes. At 100% load the device removes 75%, but at 10% load it removes only about 66% (9.0 → 3.1). Low-load modes carry real weight in the cycle, so an RECD that struggles at light load can pull the weighted result close to the line. This is common, because exhaust temperatures at light load may be too low for passive regeneration. If your DG mostly runs light, look at the per-mode numbers, not just the weighted figure.

Worked Example 3: Account for Measurement Uncertainty

Every gravimetric result carries uncertainty from filter weighing, flow measurement and sampling. Suppose the lab states ±10% on each PM result. Take the result in Example 2 and look at the worst case:

• Before, low end: 0.340 × 0.90 = 0.306 g/kWh
• After, high end: 0.093 × 1.10 = 0.102 g/kWh
• Worst-case reduction: (0.306 − 0.102) ÷ 0.306 = 66.7%

A nominal 72.6% could, within stated uncertainty, be below 70%. In practice:

• Run repeat tests, ideally at least two or three per load point, and average them
• Ask the lab to report uncertainty with the result
• In supply contracts, define whether the guarantee applies to the nominal value or the lower confidence bound. Our article on RECD performance guarantees and operating conditions covers how to word this

Don’t Forget the Other Gases

The CPCB condition also says the RECD should not adversely affect other emissions. Run the same before/after comparison for CO, HC and NOx on the same basis. Because CO and HC are often reduced by an oxidation stage, many specifications ask for these reductions too. A large NOx rise or an unexpected change in NO₂ share is worth questioning with the supplier.

---

Part 5: RECD Emissions Data in mg/Nm³ vs g/kWh, and Why Units Matter

Two reports on the same RECD can both be correct and still look impossible to compare, because one is in mg/Nm³ and the other in g/kWh. Each unit answers a different question.

mg/Nm³: How Concentrated Is the Exhaust?

Milligrams per normal cubic metre gives the mass of particulate in a standard volume of exhaust. “Normal” means the volume has been corrected to standard temperature and pressure (commonly 0 °C and 1 atm, though some reports use 25 °C, so check). This unit is:

• Natural for stack monitoring. It is what a stack-sampling lab measures directly.
• Familiar to State Pollution Control Boards, because stationary-source consent conditions are often written in it
• Sensitive to dilution. Extra air lowers the number without any change to the engine’s actual emissions. That is why the O₂ reference in Part 4 matters so much.
• Blind to engine size. 50 mg/Nm³ from a 25 kVA set and from a 2,000 kVA set are very different amounts of pollution.

g/kWh: How Much Pollution per Unit of Useful Work?

Grams per kilowatt-hour relates the particulate mass emitted to the electrical or shaft energy produced. This unit is:

• The basis of CPCB genset emission norms. Engine type approval limits under CPCB II and CPCB IV+ are expressed this way (see CPCB II vs CPCB III vs CPCB IV+ norms explained)
• Immune to dilution. Adding air doesn’t change grams emitted or kWh produced.
• Fair across sizes. It lets you compare engines of different ratings
• More work to obtain. It needs exhaust flow (or fuel flow) and an accurate power measurement as well as concentration

Converting Between the Two

The link between them is exhaust flow rate and power output:

PM (g/h) = Concentration (mg/Nm³) × Exhaust flow (Nm³/h) ÷ 1,000

PM (g/kWh) = PM (g/h) ÷ Power (kW)

Example: A stack test measures 40 mg/Nm³ (dry, at the same basis as the flow) with an exhaust flow of 450 Nm³/h while the DG delivers 60 kW.

• Mass rate: 40 × 450 ÷ 1,000 = 18 g/h
• Specific emission: 18 ÷ 60 = 0.30 g/kWh

The flow and concentration have to be on the same basis: both dry or both wet, and at the same reference conditions. Mixing a dry concentration with a wet flow can shift the answer by 5–10%. The OEM’s exhaust flow data helps as a cross-check (see using OEM exhaust flow data when specifying an RECD), but measured flow at test time is better.

When the Percentage Reduction Agrees, and When It Doesn’t

At a single load point, if both readings are properly corrected to the same O₂ reference, the percentage reduction should be about the same in either unit. The flow and power cancel out in the ratio.

The results diverge when:

• The before and after readings use different O₂ or dilution bases (as in Example 1)
• The load differed between tests, so the same mg/Nm³ represents different g/kWh
• One result is a single-mode concentration and the other a weighted multi-mode g/kWh figure. These aren’t the same measurement at all
• Back-pressure from the RECD changed engine efficiency. A significantly higher fuel burn for the same kW changes g/kWh slightly even if concentration falls

Which Unit Should Your Report Use?

Purpose	Preferred unit	Why
Comparing to CPCB engine norms	g/kWh	Same basis as the norms
Site acceptance of an RECD	Both, ideally	mg/Nm³ is measured; g/kWh proves the engine-level reduction
SPCB consent conditions	Whatever the consent letter specifies (often mg/Nm³)	Legal match to the consent
Periodic health check	Same unit as the commissioning baseline	Consistency over time

Our recommendation for site acceptance: ask the lab to report the measured concentration (mg/Nm³), the O₂/CO₂ readings, the corrected concentration at a stated reference, the exhaust flow, the measured kW and the derived g/kWh. With all six, anyone can audit the calculation, including an inspector during a Pollution Control Board inspection.

---

Bringing It Together: A Defensible RECD Test in Seven Steps

1. Define the purpose and write a one-page test plan before anyone arrives on site
2. Prepare the engine so the baseline reflects a healthy engine, not a neglected one
3. Fix the load points in measured kW and repeat them exactly
4. Measure particulate mass by an accredited lab, and use opacity only as a secondary check
5. Record O₂/CO₂, flow, temperature and back-pressure alongside PM
6. Calculate on a common basis. Correct for dilution, weight modes if needed and state uncertainty
7. Report in both mg/Nm³ and g/kWh, with enough raw data to recalculate

Following these steps gives you a result that holds up when a supplier, auditor or regulator reads it. It is also useful to keep with your other compliance papers. Our environmental compliance checklist for generator owners explains where it fits, and how to verify a CPCB-tested RECD on your DG set covers the documentation side.

---

Frequently Asked Questions

What is the minimum particulate reduction an RECD must achieve in India?

Under the CPCB framework for in-use diesel gensets, a retrofitted device must show at least 70% reduction in particulate matter without adversely affecting other emission parameters. Type testing is done by CPCB-recognised labs over the ISO 8178 D2 cycle. For the regulatory background, see why RECD is mandatory in India and the official CPCB website.

Can I test my RECD myself?

You can run your own opacity checks and log back-pressure, and both are useful for routine monitoring. Evidence of particulate reduction for compliance or contractual acceptance should come from an accredited lab using gravimetric stack sampling.

Why did my RECD test show more than 90% reduction?

High-efficiency wall-flow filters can genuinely exceed 90% by mass. Check the basis first, though. A very high figure from an opacity meter, or from uncorrected concentrations where the “after” exhaust was more diluted, may overstate the real reduction.

How often should RECD performance be re-tested?

Practice varies by state and consent conditions. A sensible approach is to log opacity and back-pressure at every service, and to repeat gravimetric testing when opacity rises, back-pressure changes noticeably, or a regulator or client asks. Check your consent letter and state-wise DG set norms.

Does running the test at full load give the best result?

It often does, because hotter exhaust helps many RECDs regenerate and perform. That is exactly why a single full-load test can flatter a device whose DG actually spends most of its life at 30–40% load. Test at the loads your set really sees.

Can the RECD supplier conduct the acceptance test?

The supplier can coordinate it. For credibility, the sampling and analysis should be done by an independent accredited lab, with the client’s representative present. Spell this out in the RFQ. Our guide on how to write a technical RFQ for an RECD project shows how.

---

Need Help Planning Your RECD Test?

Aceget supplies CPCB-type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA. We help clients plan sampling ports, load points and acceptance criteria before installation. You can view our certifications and type approvals, read up on DG set emission regulations, or contact our team with your DG nameplate details. Here’s what nameplate photos we need.