Why RECD Test Results May Differ from Day-to-Day Site Readings

The acceptance report says the Retrofit Emission Control Device removed 75% of particulate. Three weeks later, the maintenance supervisor sends a photo of a grey puff from the stack during a morning start. The EHS team’s handheld monitor in the DG room shows a spike. The quarterly stack monitoring report quotes a concentration that nobody can match to the acceptance figures. Everyone starts asking whether the RECD has stopped working.

Usually it hasn’t. Acceptance tests and day-to-day site readings are different measurements, taken under different conditions, to answer different questions. Expecting them to agree number for number is like expecting a doctor’s annual check-up result to match a fitness tracker reading taken while climbing stairs.

This article explains why RECD test results and everyday site readings differ, what each common site reading actually tells you, and how to link the two so your routine checks catch real problems without false alarms.

The Core Difference: Controlled Snapshot vs Everyday Reality

An acceptance or periodic test is a controlled snapshot:

• Load held steady at agreed points
• Engine and device fully warmed up and stable
• Regeneration excluded, or recorded separately
• Gravimetric particulate sampling by an accredited lab
• Results corrected to a defined basis
• A reduction calculated from matched before and after measurements

Day-to-day operation is everything else:

• Cold starts, short runs and weekly test runs
• Loads that change minute by minute
• Regenerations whenever the device needs them
• Seasonal temperature swings, fuel deliveries and engine wear
• Quick instruments, such as opacity meters, gauges and eyes, used without lab controls

Both views are valid. They just aren’t measuring the same thing. For background on test conditions, see planning an RECD emissions test before and after installation.

Five Site Readings, and What Each Really Tells You

1. “I saw smoke from the stack”

What it shows: a visible plume at one moment.

Why it differs from the test: Visible plumes are dominated by water vapour, condensed hydrocarbons and light-scattering particles, not necessarily soot mass. During a cold start, condensation and unburnt fuel form white or grey vapour even when the filter is working. During regeneration, released material can cause brief haze. The acceptance test deliberately excluded both conditions.

Useful for: spotting persistent black smoke under steady load, which is a genuine warning sign. See why is my generator emitting black smoke and why is my generator emitting white smoke.

Not useful for: judging particulate reduction.

2. Handheld opacity meter readings

What it shows: how much light the exhaust absorbs or scatters, as % opacity or a k-value.

Why it differs from the test: Opacity isn’t mass. It responds strongly to soot and also to water droplets and condensed organics. After an effective filter, opacity often sits near the instrument’s lower limit, so small fluctuations look large in relative terms. Readings also change with load, temperature and how long the engine has been running.

Useful for: trend monitoring at a fixed, repeatable condition, and quickly detecting a damaged filter.

Not useful for: calculating a mass reduction percentage. See smoke opacity vs particulate mass.

3. Air quality monitor in or near the DG room

What it shows: particulate concentration in the room or surrounding air, in µg/m³.

Why it differs from the test: It measures ambient air, not exhaust. Readings depend on ventilation, outdoor pollution, nearby traffic, construction dust and any exhaust leaks into the room. In Delhi NCR winters, background PM2.5 alone can be very high. See why Delhi AQI increases in winter.

Useful for: finding exhaust leaks into occupied spaces and checking room ventilation.

Not useful for: verifying the RECD. See can a handheld AQI monitor verify RECD performance.

4. Differential pressure (ΔP) gauge across the RECD

What it shows: the pressure drop across the device, which reflects soot and ash load.

Why it differs from the test: ΔP isn’t an emission reading at all. It changes with exhaust flow (and so with load), with temperature, and with soot and ash load. A reading during a light-load test run means something quite different from one at full load.

Useful for: tracking soot and ash build-up, confirming regeneration, and detecting cracks or bypass leaks (an unexpected drop). This is one of the most valuable routine readings available.

Not useful for: stating a reduction percentage directly.

5. Periodic stack monitoring report for consent compliance

What it shows: stack concentrations, often PM, CO, NOx and SO₂, measured by a lab at the time of the visit.

Why it differs from the acceptance test: Consent monitoring usually measures only downstream concentration, at whatever load the site needs that day, often on a different basis (wet or dry, reference O₂ or actual) from the acceptance test. It doesn’t measure a baseline, so it can’t produce a reduction percentage. See how to compare RECD test results from different laboratories.

Useful for: demonstrating compliance with stack limits in your consent, and as a long-term trend if conditions are kept similar.

Not useful for: proving the RECD’s percentage reduction.

Reading-by-Reading Summary

Site reading	Measures	Typical reasons it varies	Best use
Visual smoke	Plume appearance	Cold start, regeneration, humidity, light	Spot persistent black smoke
Opacity meter	Light extinction	Load, warm-up, water vapour, instrument floor	Trend at fixed condition; damage detection
AQI monitor	Ambient PM	Ventilation, background pollution, leaks	Room air quality; leak detection
ΔP gauge	Pressure drop across RECD	Load, temperature, soot and ash	Soot/ash trend; regeneration; crack detection
Consent stack report	Downstream concentration	Load, basis, season, engine state	Compliance with stack limits

The Conditions Behind the Difference

Even when two measurements use the same principle, operating conditions push them apart. The most common reasons a day-to-day reading differs from the test:

Load

Site loads vary throughout the day. Emissions, exhaust temperature, ΔP and device efficiency all change with load. See why generator load must be matched in RECD comparison tests.

Warm-up and short runs

Many DG sets run only briefly, for weekly tests or short outages, and never reach the thermal stability of an acceptance test. See how to account for warm-up time when testing an RECD.

Regeneration

Regeneration events are excluded from steady-state acceptance tests but happen during normal operation. During and just after them, downstream readings rise and ΔP falls. See RECD testing during regeneration.

Device age

Soot cake, ash and catalyst ageing change the device’s behaviour over time. See can RECD efficiency change between new and aged conditions.

Engine and fuel condition

A new fuel delivery, a clogged air filter or worn injectors change engine-out emissions and therefore what the RECD receives. See how fuel and engine condition can confound RECD performance tests.

Season and ambient conditions

Hot summer intake air, humid monsoon conditions and cold winter starts all affect combustion and condensation in different ways.

Worked Example: One Week of Site Readings

The figures are illustrative, for a 200 kVA set whose acceptance test showed a 75% PM reduction at 50% load. The site logs opacity and ΔP at every run.

Day	Run type	Load	Duration	Opacity (%)	ΔP (mbar)	Note
Mon	Weekly test run	20%	15 min	4.8	22	Cold, short run, white vapour at start
Tue	Grid outage	55%	2 h	0.7	46	Warm, steady: close to test conditions
Wed	Grid outage	70%	45 min	0.9	58	Higher load, higher ΔP
Thu	Grid outage	55%	1.5 h	2.6	31	ΔP fell mid-run; controller logged regeneration
Fri	Grid outage	55%	2 h	0.8	45	Back to normal

Monday’s 4.8% opacity alarmed the supervisor, but it came from a cold, short, light-load run, nothing like the acceptance test. Thursday’s rise came with a regeneration event, confirmed by the falling ΔP and the controller log. Tuesday and Friday, the runs closest to test conditions, show consistent, low readings.

Conclusion: the RECD is behaving normally. The readings differ from the test because the conditions differ.

Fingerprint Your RECD on Test Day

The best way to connect test results with site readings is to record a fingerprint of the device during the acceptance or periodic test, using the same simple instruments your site team will use afterwards.

During the lab test, at each load point, also record:

• Opacity, with your site’s opacity meter, at the downstream port or stack outlet
• ΔP across the RECD, with the site’s gauge
• Engine back-pressure
• Exhaust temperature before and after the RECD
• Load in kW

These readings become your reference values. Later site readings taken under the same conditions (same load, fully warm, no regeneration) can be compared directly with them.

Example fingerprint and trigger levels

The figures are illustrative.

Indicator (at 50% load, warm, no regeneration)	Reference at acceptance	Investigate if	Act urgently if
Downstream opacity	0.7%	> 2.0% on two consecutive checks	> 5.0%
ΔP across RECD	44 mbar	> 65 mbar (ash/soot build-up)	< 25 mbar without recent cleaning (possible crack or bypass)
Engine back-pressure	Within OEM limit	Within 10% of limit	Above limit
Exhaust temperature rise across RECD	+5 °C	Persistent large rise outside regeneration	—

Set trigger values with your supplier, based on the device’s design and the fingerprint. Then a site reading has a clear meaning: normal, investigate or act. That replaces guesswork about whether a number “looks bad.”

Building a Routine That Works

1. Pick one standard check condition. For example, 50% load, at least 45 minutes into a run, no regeneration active. Use it for every routine reading.
2. Use the same instruments every time, calibrated and identified by serial number. See which instrument calibration records should accompany RECD tests.
3. Log readings against operating hours, not just dates, so ageing trends are visible.
4. Record context: fuel delivery dates, maintenance work, regeneration events and ambient temperature.
5. Review trends monthly, not just individual readings.
6. Escalate based on agreed triggers, not on one surprising number.

This fits naturally into a monthly maintenance checklist for diesel generators and the RECD maintenance guide.

Seasonal Patterns to Expect

Many sites see predictable seasonal differences in their routine readings. Knowing the pattern prevents false alarms.

Season (north India)	Typical effect on site readings	Why
Winter (Dec–Feb)	More white vapour at start-up; longer warm-up; ΔP lower at cold start	Cold ambient and condensation; denser intake air
Spring (Mar–Apr)	Readings close to reference values	Moderate temperatures
Summer (May–Jun)	Slightly higher opacity at high load; higher exhaust temperatures; more frequent passive regeneration	Hot, less dense intake air; heavier loads from cooling
Monsoon (Jul–Sep)	Water ingress risk; more vapour; possible corrosion after idle periods	High humidity; rain entering stacks
Post-monsoon (Oct–Nov)	Higher DG-room AQI readings, unrelated to the RECD	Rising background pollution; GRAP season begins

Comparing this month’s readings with the same month last year is often more informative than comparing with last month. See why Delhi AQI increases in winter and how GRAP restrictions affect businesses using DG sets.

A Decision Path for an Unexpected Reading

When a site reading looks wrong, work through these steps before concluding the RECD has a problem:

1. Check the conditions. Was the reading taken at the standard check condition: correct load, fully warm, no regeneration? If not, repeat it at the standard condition.
2. Check the instrument. Is it the usual instrument, calibrated, zeroed and used the same way? See which instrument calibration records should accompany RECD tests.
3. Check for regeneration. Look at the controller log and ΔP trend. Was the device regenerating or just after a regeneration?
4. Check for recent changes. Fuel delivery, maintenance work, exhaust modifications or a change in the load connected to the set.
5. Repeat the reading at the standard condition on the next run.
6. Compare with trigger values. If the repeated reading is still outside the “investigate” band, inspect the device and exhaust joints.
7. Escalate if needed. If inspection doesn’t explain it, or readings reach the “act urgently” band, contact the supplier and consider a retest. See when should an RECD be retested after changes to the DG set.

Most unexpected readings resolve at steps 1 to 3. Documenting each step also gives you a clear record if a regulator or auditor asks about a reading later.

What to Record With Every Site Reading

A site reading is only useful if you know the conditions it was taken under. A simple log line for each check should include:

Field	Why
Date, time and DG hour meter reading	Links the reading to operating hours and ageing
Load in kW (not kVA)	Readings change with load
Minutes since start	Shows whether the set was fully warm
Exhaust temperature before and after the RECD	Confirms thermal state; reveals regeneration
ΔP across the RECD	Soot/ash load and damage detection
Downstream opacity	Trend and damage detection
Engine back-pressure	Engine protection
Regeneration status from controller	Explains spikes and ΔP changes
Ambient temperature	Seasonal context
Instrument ID and operator	Traceability
Notes (fuel delivery, maintenance, visible smoke)	Context for unusual readings

A log like this takes a couple of minutes per check, and it turns scattered numbers into evidence.

When a Difference Does Mean a Problem

Not every difference is harmless. These patterns, seen at the standard check condition, deserve investigation:

• Opacity rising over several checks while load and warm-up are unchanged
• ΔP dropping sharply without cleaning or regeneration, which suggests a crack, failed seal or bypass
• ΔP rising steadily and regeneration not bringing it down, which suggests ash build-up or failed regeneration
• Back-pressure approaching the engine maker’s limit
• Persistent black smoke at steady, warm load
• A consent stack report clearly worse than previous reports taken under similar load and basis

If these appear, inspect the device and the exhaust connections, then consider a retest. See when should an RECD be retested after changes to the DG set.

Explaining the Difference to Others

Regulators, neighbours and management may see a smoky start or a high room reading and conclude the RECD has failed. A short, factual explanation helps:

“The RECD’s performance was measured under controlled conditions by an accredited laboratory on [date], showing a [X]% particulate reduction. Day-to-day readings vary with load, start-up and the device’s normal cleaning cycle. We monitor the device weekly against reference values set during that test, and the latest readings are within the normal range.”

Keep the acceptance report, fingerprint values and recent trend log together, ready for a Pollution Control Board inspection.

Related Reading

• RECD particle number vs particle mass: what each test shows
• How to interpret uncertainty in an RECD emission reduction claim
• Can RECD efficiency change between new and aged conditions?
• When should an RECD be retested after changes to the DG set?

Frequently Asked Questions

Why does my stack look smoky at start-up if the RECD works?

At start-up, the engine and device are cold. Water vapour and unburnt hydrocarbons condense into visible white or grey vapour, and the catalyst hasn’t reached working temperature. This usually clears as the set warms up. Persistent black smoke at steady, warm load is the real warning sign.

Can I calculate the RECD’s efficiency from my opacity meter?

No. Opacity isn’t mass, and after an effective filter it is often near the instrument’s lower limit. Use opacity for trends and damage detection at a fixed condition, not for efficiency figures.

My consent stack report shows a different number from the acceptance test. Is that a problem?

Usually not. Consent monitoring typically measures only downstream concentration, at whatever load the site needs, often on a different basis, and without a baseline. Compare it with previous consent reports taken under similar conditions, not with the acceptance reduction percentage.

What single routine reading is most useful?

Differential pressure across the RECD at a standard load and warm-up condition. It tracks soot and ash build-up, confirms regeneration and can reveal cracks or bypass leaks.

Should I set my own trigger values or ask the supplier?

Set them together. The supplier knows the device’s design limits for back-pressure, ash loading and normal ΔP ranges. You know your site’s duty cycle and how readings vary in practice. Start with values derived from the acceptance-test fingerprint and the supplier’s guidance, then refine them after a few months of routine data. Record the agreed values in writing, so everyone on the site team applies the same thresholds.

How often should I take routine readings?

At every planned run, such as weekly tests, if practical, and at least monthly at the standard check condition. More frequent readings make trends easier to see.

Link Your Site Checks to Your Test Results

Lab tests and site readings each have a job. When the test also fingerprints the device with the same instruments your site uses, routine readings become meaningful instead of worrying. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA, and we help clients set reference values and trigger levels at commissioning. See our certifications, read about DG set emission regulations, or contact our team.