How Exhaust Sampling Location Affects RECD Test Results

A particulate sample represents one small stream of gas, drawn through one small nozzle, at one point in a pipe. The lab then treats that sample as if it described all the exhaust leaving the generator. Whether that assumption holds depends almost entirely on where the nozzle sits.

Put it in the wrong place and you can get a result that is precise, repeatable, well documented, and still wrong. Put the “before” port in one kind of location and the “after” port in another, and you can manufacture a reduction, or hide one, without the Retrofit Emission Control Device doing anything different.

Instead of listing rules in the abstract, this article walks along a typical diesel generator exhaust from the engine to the top of the chimney, stopping at each point where someone might be tempted to drill a sampling port. At each stop it asks what the gas is doing there and whether it’s a good place to measure.

Why Location Matters: Three Things a Sample Must Be

For a sample to stand in for the whole exhaust, three conditions have to hold:

1. Representative. Particles and gases are spread evenly across the pipe, or the lab samples enough points to average out any unevenness.
2. Unaltered. Nothing between the source and the probe has added, removed or changed the material being measured, for example by leaks, deposits, condensation or other gensets.
3. Comparable. The “before” and “after” samples were taken under equivalent conditions, so the only meaningful difference between them is the RECD.

Every location decision affects at least one of these three.

Stop 1: The Exhaust Manifold and Turbocharger Outlet

What the gas is doing: It’s at its hottest, often 400–550 °C or more at high load, with high velocity and strong pulsations from individual cylinders. After a turbocharger the flow swirls strongly.

Is it a good sampling point? Rarely. The pulsing, swirling flow makes isokinetic sampling (matching the probe’s intake speed to the gas speed) very difficult. Temperatures are hard on sampling equipment. There’s seldom room or safe access for a lab’s probe.

What it’s good for: temperature and back-pressure measurement. A tapping here is the right place to monitor engine back-pressure, which you’ll want to log during any RECD test. See how exhaust back-pressure affects a DG set.

Stop 2: The Flexible Bellows

What the gas is doing: Passing through a corrugated section that absorbs engine vibration and thermal movement. The corrugations disturb flow, and bellows are a common place for small leaks to develop.

Is it a good sampling point? No. You can’t fit a port into a bellows, and the disturbed flow just after it isn’t representative.

Why it still matters: a leaking bellows downstream of the RECD but upstream of the sampling port lets ambient air into the exhaust. That dilutes the sample and can make the device look better than it is. Inspect bellows and flanges before testing. For layout guidance, see flexible bellows in an RECD exhaust system. Dilution is covered in depth in how dilution can mislead RECD before-and-after measurements.

Stop 3: The Straight Run Just Before the RECD

What the gas is doing: If there is a decent straight length here, the flow has had a chance to settle after the bellows and any bends. The gas is still hot and hasn’t yet passed through the device.

Is it a good sampling point? Often the best available upstream port for an upstream/downstream comparison, provided:

• There is enough straight pipe on both sides of the port
• There is no bend, reducer or tee immediately before it
• The pipe diameter is the same as, or documented relative to, the downstream port

What to watch: the RECD’s inlet cone. As the pipe widens into the device housing, the flow decelerates and can separate from the walls. A port too close to the inlet cone sees this disturbed flow.

Stop 4: Inside and Immediately After the RECD Outlet Cone

What the gas is doing: It has just passed through the substrate, which straightens and evens the flow to some extent. Then it hits the outlet cone, where the flow converges back into the pipe and can form jets and eddies. Temperature is lower than at the inlet if the device is still warming, or higher during active regeneration.

Is it a good sampling point? Not immediately after the cone. Give the flow a few pipe diameters to reorganise. Ports fitted directly onto the RECD outlet flange for convenience are a common source of unrepresentative downstream samples.

Related: the RECD’s position along the exhaust route affects how much straight pipe is available on either side. Our guides on where to position an RECD along a long exhaust run and what to include in an RECD general arrangement drawing cover planning port locations at the design stage.

Stop 5: The Silencer

What the gas is doing: Passing through chambers, baffles or absorption packing designed to reduce noise. These change flow direction repeatedly, and they can trap and later release soot.

Is it a good sampling point? Neither inside nor just after it. Two specific risks:

• Deposit release. Old silencers accumulate soot from years of operation without an RECD. When flow or temperature changes, that soot can break loose. A downstream port after the silencer may record old deposits, not current emissions.
• Flow disturbance. The outlet of a silencer is seldom a clean, even flow.

If the RECD replaces the silencer or shares a frame with it, the layout changes. See can an RECD replace a generator’s exhaust silencer and RECD on a shared support frame with an existing silencer.

Stop 6: Horizontal Runs and Bends

What the gas is doing: In long horizontal runs, especially oversized ones, the heavier particles can settle slightly and the flow can stratify, particularly at low velocity. At every bend the flow is thrown to the outside wall and sets up secondary swirl that persists for several diameters.

Is it a good sampling point? A long straight horizontal run can be acceptable if the velocity is reasonable and the port is well clear of bends. Ports just after an elbow are a poor choice. Our review of bends and reducers in RECD exhaust layouts shows how to keep the route clean.

Stop 7: The Vertical Stack, Mid-Height

What the gas is doing: If the vertical section is long and straight, the flow has settled. The gas has cooled further. On a tall chimney, that may be well below the temperature at the RECD.

Is it a good sampling point? Often the best available downstream port, which is why many stack-monitoring ports are fitted here. It meets the “representative” test well. Its weakness is “comparable”: if the upstream port is near the engine and the downstream port is high up a cool chimney, the two samples see gas at very different temperatures, which affects condensable particulate (see below).

Access: a good port is useless if the sampling team can’t reach it safely. Plan platforms and ladders early. See safe access to RECD sampling ports on a tall chimney.

Stop 8: The Stack Top and Rain Cap

What the gas is doing: Leaving the stack, where wind across the opening can cause pressure fluctuations, and in gusty conditions ambient air can be drawn a short way back down into the stack. Rain caps and flaps add further disturbance.

Is it a good sampling point? No. Ports within the last few diameters of the outlet risk sampling diluted, unsteady gas. Measurements taken outside the stack, in the plume, are a different exercise altogether and can’t verify an RECD. We explain why in can a handheld AQI monitor verify RECD performance.

The Rules Behind the Walk

Rule 1: Straight lengths and distance from disturbances

Stationary-source sampling practice deals with disturbed flow by placing ports well away from bends, expansions, contractions and fittings. The widely used reference is US EPA Method 1. It sets out an ideal of at least eight pipe diameters downstream and two diameters upstream of the nearest disturbance, and requires more sampling (traverse) points when shorter straight lengths are unavoidable. Indian stack monitoring follows the same principles.

On real DG installations, eight clear diameters are often impossible, especially in basements and plant rooms. When you can’t meet the ideal:

• Choose the location with the longest available straight run
• Record the actual upstream and downstream distances in pipe diameters on the test report
• Have the lab carry out a velocity traverse and sample at multiple points across the pipe instead of one centre point

Rule 2: Isokinetic sampling

Particles have inertia. If the probe draws gas faster than the surrounding flow, it pulls in extra gas but misses some of the larger particles, which carry straight past. If it draws slower, larger particles are over-collected. Isokinetic sampling, where nozzle velocity equals local gas velocity, avoids this bias. It needs a stable, measurable velocity at the port, which is another reason to keep ports away from disturbances. The particulate sampling approach used in stack testing follows the same principles as US EPA Method 5.

Rule 3: Temperature at the port affects what counts as particulate

Part of diesel particulate is condensable. Some organic compounds and sulfuric acid exist as vapour in hot exhaust and only become particles as the gas cools. A port where the gas is at 400 °C and a port where it is at 150 °C can collect different PM mass from identical exhaust, depending on how the sampling train handles temperature.

This matters most when the upstream and downstream ports sit at very different temperatures, for example upstream near the engine and downstream at the top of a long uninsulated chimney. Part of the apparent “reduction” may then be a change in the condensable fraction, not the RECD’s filtration.

What to do:

• Record gas temperature at every port during every sample
• Make sure the lab uses the same probe and filter temperature settings for both ports
• Where possible, choose ports at similar gas temperatures, or use a dilution-based method that standardises the cooling step

Rule 4: Nothing between the device and the port

Between the RECD outlet and the downstream port, there should be:

• No unsealed flanges or leaking bellows (air ingress dilutes the sample)
• No connection from another genset (other exhaust contaminates the sample)
• No drain points open to atmosphere
• No silencer that could release old deposits, if it can be avoided

Special Case: Shared Chimneys

Many sites route two or more DG sets into one chimney. A downstream port on the common section sees exhaust from every set that is running.

• During testing, isolate the other sets, or sample before the junction
• Watch for backflow. With one set off, exhaust from a running set can enter the idle set’s branch if there are no isolation dampers, and leak out elsewhere
• If one RECD is shared by two sets, the test must reflect how they actually operate. See can two diesel generators share one RECD and RECD fitment on DG sets with multiple exhaust outlets

Special Case: Sequential Tests and Changing Pipework

In a sequential test, the baseline is taken on the original exhaust and the post-RECD test after installation. If installing the RECD changed the pipework (moving bends, adding reducers, shortening straight runs), the original port may no longer be representative, or may no longer exist.

Before signing off a sequential comparison, confirm that:

• The downstream port is at the same physical location as the baseline port, or the change is documented
• The straight lengths around the port haven’t been shortened
• The pipe diameter at the port is unchanged

If the layout has changed substantially, consider an upstream/downstream test instead. It compares the device against itself on the same day, which avoids the issue entirely. For more on choosing a comparison method, see planning an RECD emissions test before and after installation. If you are reusing existing exhaust pipework for the retrofit, plan port positions as part of that review.

Port Hardware: Small Details That Cause Big Problems

The port itself matters as much as where it sits:

• Size: large enough for the lab’s isokinetic probe and pitot tube. Ask the lab before fabrication.
• Orientation: for horizontal pipes, ports at the side allow traverses across the diameter. For vertical stacks, two ports at 90° to each other allow a full traverse.
• Sealing: a threaded cap or flange that seals properly when not in use. An open or loose port is itself an air leak.
• Projection: a port nozzle that sticks too far into the flow disturbs it. One that is too short may not let the probe reach the centre.
• Insulation: port fittings on insulated pipe should be accessible without tearing away insulation every time.

Pairing Upstream and Downstream Ports

In an upstream/downstream comparison, aim for “twin” ports that differ only in being on either side of the RECD:

Characteristic	Upstream port	Downstream port	Goal
Pipe diameter	D	D	Same
Straight run before port	≥ several D	≥ several D	Similar, documented
Straight run after port	≥ 2 D	≥ 2 D	Similar, documented
Gas temperature	T₁	T₂	As close as practical, recorded
Leaks between source and port	None	None	Verified by inspection
Other gensets connected	No	No	Verified

Where the twins can’t be perfect, document the differences. A report that says “the downstream port is 3D after the RECD outlet cone; the upstream port is 6D after the bellows” lets a reviewer weigh the result. A report that says “sampling port” tells them nothing.

Worked Example: Same Exhaust, Two Downstream Ports

The figures are illustrative. A 250 kVA set in a commercial building has its RECD in the plant room. After the RECD, the exhaust passes through an old silencer, along a horizontal run with two bends, and up a 20-metre chimney. There are two possible downstream ports:

• Port A: an existing stack-monitoring port near the top of the chimney, about 2 diameters below the rain cap, after the silencer
• Port B: a new port fitted 5 diameters after the RECD outlet cone, before the silencer

The upstream port, before the RECD, gives 72 mg/Nm³ at a matched 150 kW. Both downstream ports are sampled the same afternoon, using the same equipment settings:

	Port A (stack top)	Port B (after RECD)
Gas temperature at port	185 °C	390 °C
CO₂ at port (%)	7.1	8.4
Upstream CO₂ (%)	8.4	8.4
PM measured (mg/Nm³)	19.5	20.2
PM corrected to upstream CO₂ basis	23.1	20.2
Apparent reduction (uncorrected)	72.9%	71.9%
Reduction on a common CO₂ basis	67.9%	71.9%

On the uncorrected numbers, the two ports agree to within a point, and both pass. The CO₂ readings tell a different story. Port A is drawing in ambient air near the stack top, so its sample is diluted. Once that is corrected, Port A shows more particulate than Port B. The extra mass is most likely old soot released from the silencer and bends downstream of the device, which says nothing about the RECD’s own performance.

Port B is the better choice. It is close enough to the device to avoid the silencer and the stack-top effects, and far enough past the outlet cone for the flow to settle. Port A is still useful for routine stack monitoring, but it shouldn’t be used to judge the RECD.

Port Location Checklist

• Ports shown on the general arrangement drawing before fabrication
• Straight lengths before and after each port recorded in pipe diameters
• No bends, reducers, bellows or cones immediately before either port
• No unsealed joints between the RECD and the downstream port
• Other gensets on shared stacks isolated during testing
• Gas temperature recorded at each port during each sample
• Same probe and filter temperature settings used at both ports
• Port size, orientation and caps suitable for the lab’s equipment
• Safe access confirmed for the sampling team
• For sequential tests: port location unchanged between baseline and post-RECD test

How Sampling Location Links to the Other Test Variables

• Load changes exhaust velocity, which affects isokinetic sampling and how evenly particles spread. See why generator load must be matched in RECD comparison tests.
• Warm-up determines the gas temperature at each port. See how to account for warm-up time when testing an RECD.
• Regeneration spikes are best captured by a well-placed downstream port. See RECD testing during regeneration.
• Dilution from leaks depends directly on what lies between the device and the port. See how dilution can mislead RECD before-and-after measurements.

Frequently Asked Questions

Can I use my existing annual stack-monitoring port for RECD testing?

Possibly. Check that it has adequate straight lengths, that nothing has changed between it and the RECD, that it isn’t near the stack top, and that it isn’t on a section shared with another genset. Many older ports were placed for convenience, not representativeness.

What if there is no room for eight diameters of straight pipe?

That’s common on DG installations. Choose the longest straight run available, record the actual distances, and ask the lab to sample at multiple traverse points. The result is still usable, as long as the limitation is documented and the same approach is used in both tests.

Should the downstream port be right after the RECD to avoid leaks?

Close enough to avoid long runs with joints, but far enough past the outlet cone for the flow to settle. A few pipe diameters after the cone is a reasonable compromise when space is tight.

Does the port location matter if I’m only using an opacity meter?

Yes. Opacity meters also need representative, undiluted exhaust. But remember that opacity alone can’t prove a particulate mass reduction. See smoke opacity vs particulate mass.

Who decides where the sampling ports go?

Ideally, the RECD supplier, the installer and the testing lab agree on them together during design. Write the agreed locations into the purchase order and general arrangement drawing so they’re built correctly the first time. Our guide on customer and installer responsibilities in an RECD project explains who usually owns which task.

Put the Ports in the Right Place the First Time

Drilling a new sampling port after the exhaust is insulated and commissioned is expensive and disruptive. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA, and we plan sampling port positions, straight lengths and access with clients at the design stage. Choose a lab accredited by NABL for stack emission monitoring, review our certifications, and contact our team with your exhaust layout drawings.