- October 16, 2026
- Posted by: editor
- Category: RECD & Emission Control Devices
A performance test that only the supplier saw is easy to doubt. A performance test that the buyer, the supplier, an independent lab and perhaps a consultant all watched, according to a protocol they agreed beforehand, is much harder to dispute.
That is the point of a witnessed demonstration. It’s not mainly about the instruments. It’s about agreement: who decided the conditions, who saw the samples being taken, who confirmed the numbers, and who signed off before anyone left the site.
This guide follows the timeline of a witnessed RECD performance demonstration, from the weeks before testing to the day the final report is accepted. It covers the protocol, the people, the hold points where testing pauses for agreement, chain of custody for filter samples, the test-day agenda, and the sign-off.
When a Witnessed Demonstration Is Worth the Effort
Not every RECD installation needs a formal witnessed test. Consider one when:
- A payment milestone depends on the result (see RECD payment milestones and project deliverables)
- The supply contract includes a performance guarantee (see RECD performance guarantees and operating conditions)
- The installation is a trial before a larger rollout (see RECD trial installation acceptance criteria)
- The site is high-profile or regulated, such as a hospital, data centre or pharmaceutical plant
- A previous result has been disputed, or two labs have disagreed (see how to compare RECD test results from different laboratories)
- A corporate, lender or ESG audit will rely on the figure
For a single small DG set with no contractual consequences, a standard accredited lab test with good documentation may be enough.
4–6 Weeks Before: Agree the Protocol
The protocol is the most important document. It should be agreed and signed before anyone books a lab or a load bank. A practical protocol covers these headings.
1. Purpose and acceptance criterion
State exactly what is being demonstrated. For example: “particulate reduction of at least 70% by mass at 50% and 75% of rated kW, measured upstream and downstream of the RECD on the same day, using the average of two valid runs at each load point.”
Name the pass criterion and how it will be calculated, including whether measurement uncertainty is considered. See how to calculate RECD particulate reduction.
2. Test design
- Comparison method: upstream/downstream on the same day is the strongest choice for a witnessed test
- Port locations, with straight lengths documented (see how exhaust sampling location affects RECD test results)
- Number of runs per load point and what counts as a valid run
3. Operating conditions
- Load points in kW, tolerance, and load source (see why generator load must be matched in RECD comparison tests)
- Stability criterion before sampling (see how to account for warm-up time when testing an RECD)
- How regeneration will be handled (see RECD testing during regeneration)
- Engine and fuel condition requirements (see how fuel and engine condition can confound RECD performance tests)
4. Measurement methods and reporting
- Sampling methods for particulate and gases
- Reporting basis: wet or dry, reference O₂, reference temperature and pressure, and units (mg/Nm³ and g/kWh)
- Required calibration annex (see which instrument calibration records should accompany RECD tests)
- Uncertainty reporting
5. Roles, witnesses and decision-making
Who attends, what each person is responsible for, and who can make decisions on the day (see the roles section below).
6. Deviations and contingencies
What happens if something goes wrong: load can’t be held, a regeneration starts mid-sample, an instrument fails, or the weather prevents safe access. Agree in advance whether a run is voided, repeated or the test is rescheduled.
7. Safety
Permits to work, hot-surface precautions, safe access to sampling ports, electrical isolation for load bank connections, and emergency contacts. See safe access to RECD sampling ports on a tall chimney.
Our guides on writing measurable performance requirements into RECD tenders and customer and installer responsibilities in an RECD project include wording that can be adapted into the protocol.
Who Should Be There, and What They Do
| Role | Typically | Responsibilities on the day |
|---|---|---|
| Test coordinator | Buyer’s project engineer or appointed consultant | Runs the agenda, keeps the hold-point log, records decisions |
| Buyer’s witness | Facility or EHS manager | Observes, confirms conditions, signs at hold points |
| Supplier’s representative | RECD supplier’s engineer | Confirms device readiness, answers technical questions, signs |
| Independent laboratory | Lab accredited by NABL for stack emission monitoring | Performs sampling and analysis, follows the method, signs field records |
| DG operator | Site or AMC technician | Starts and loads the set, holds load, logs engine data |
| Load bank operator (if used) | Hire company technician | Sets and holds load, logs kW and PF |
| Safety officer | Site EHS | Permits, access checks, emergency arrangements |
| Optional third-party witness | Consultant, lender’s engineer, auditor | Independent observation |
Keep the group small enough to work safely around a hot exhaust system. Too many observers crowding a platform is a safety risk, not extra assurance.
One rule to agree in advance: only the test coordinator gives instructions to the DG operator and the lab during the test. Witnesses raise concerns with the coordinator, not directly with the people running the test.
1–2 Weeks Before: Prepare and Pre-Check
- Confirm the lab’s scope and calibration pack. Ask for instrument lists and certificates in advance, so problems can be fixed before the test day.
- Inspect the sampling ports. Check sizes, caps, access platforms and straight lengths. Fix any leaks in the exhaust between the RECD and the downstream port.
- Engine readiness. Complete any planned service at least several running hours before the test, then freeze the engine (no further adjustments). Record its condition.
- Fuel. Arrange enough fuel from one batch for the whole test, and keep the supplier certificate.
- Load bank. Confirm capacity, cable runs, safe placement and connection method.
- Device readiness. Confirm with the supplier that the RECD has the agreed operating hours since installation, and check its differential pressure and regeneration history.
- Dry run (optional but valuable). Run the DG through the load sequence a few days beforehand to confirm load stability and estimate warm-up time.
Our RECD delivery inspection checklist and guide to closing the RECD punch list before handover help make sure the installation itself is complete before it’s tested.
The Day Before: Final Confirmations
- Weather forecast checked for safe access, especially for rooftop or tall-stack ports
- Permits to work approved
- Test agenda circulated to all attendees, with arrival times
- Lab confirms filters are pre-weighed and logged, with sample IDs assigned
- DG operator confirms fuel level, battery and readiness
- Coordinator prepares the hold-point log and sign-off sheets
Hold Points: Where the Test Pauses for Agreement
A hold point is a moment where testing stops until the witnesses confirm a condition. Hold points turn “we assumed it was fine” into “we all checked it was fine.” A typical set for an RECD demonstration:
| # | Hold point | What is confirmed | Signed by |
|---|---|---|---|
| H1 | Pre-start inspection | Port locations, leak-free exhaust, instruments match the calibration pack, fuel batch | All parties |
| H2 | Lab set-up complete | Pre-test leak checks passed with recorded values; analyser zero/span checks recorded | Lab, coordinator, buyer’s witness |
| H3 | Load point 1 stable | Load within tolerance; stability criterion met; no regeneration active | Coordinator, buyer, supplier |
| H4 | Load point 1 runs complete | Runs valid (isokinetic ratio, leak checks, load log); filters sealed and logged | Lab, coordinator |
| H5 | Load point 2 stable | As H3 | Coordinator, buyer, supplier |
| H6 | Load point 2 runs complete | As H4 | Lab, coordinator |
| H7 | Post-test checks | Post-test leak checks and analyser drift recorded; differential pressure and engine data captured | Lab, coordinator |
| H8 | Field record sign-off | Field data sheets, load logs and chain-of-custody forms signed by all parties | All parties |
Between hold points, the work is witness points: witnesses may observe, but testing doesn’t stop for them.
Chain of Custody for Particulate Filters
In a gravimetric test, the filters are the evidence. If anyone could question whether a filter was swapped, contaminated or mislabelled, the result is weakened. Chain of custody is how you prevent that.
Before the test
- Each filter is pre-weighed at the lab and assigned a unique ID
- Pre-weights are recorded and, ideally, shared with the coordinator before the test day
- Filters travel in sealed, labelled holders
During the test
- The witness sees each filter holder opened and loaded, and notes the ID
- Each sample’s ID, port, load point, start and end times are recorded on the field sheet
- After sampling, the filter is sealed in its holder with a tamper-evident seal or tape signed by the lab and a witness
- Field blanks are handled the same way, but not exposed to exhaust
After the test
- A custody form lists every filter ID, seal number and signature, and travels with the samples
- The lab records the condition of each seal on receipt
- Post-test weighing follows the lab’s conditioning procedure. Witnesses may attend weighing if the protocol allows
- The report lists every filter ID with its pre-weight, post-weight and mass gain
A simple custody form looks like this:
| Filter ID | Port | Load point | Run | Start | End | Seal no. | Lab sign | Witness sign |
|---|---|---|---|---|---|---|---|---|
| F-0412 | Upstream | 75% | 1 | 10:42 | 11:12 | S-1187 | ||
| F-0413 | Downstream | 75% | 1 | 10:42 | 11:12 | S-1188 | ||
| F-0420 | Field blank | — | — | — | — | S-1195 |
Test-Day Agenda: An Example
The times below are an example for a mid-size DG set with an upstream/downstream test at two load points and two runs each. Adjust them to your site and warm-up behaviour.
| Time | Activity | Hold point |
|---|---|---|
| 08:00 | Arrival, safety briefing, permits | — |
| 08:30 | Pre-start inspection of ports, exhaust, instruments, fuel | H1 |
| 09:00 | Lab set-up, leak checks, analyser zero/span | H2 |
| 09:30 | DG start; load to 75% of rated kW; warm-up logging begins | — |
| 10:30 | Stability confirmed at 75%; regeneration status checked | H3 |
| 10:40 | Run 1 at 75% (upstream and downstream) | — |
| 11:20 | Run 2 at 75% | — |
| 12:00 | Runs reviewed; filters sealed | H4 |
| 12:10 | Step down to 50%; re-stabilise | — |
| 13:00 | Stability confirmed at 50% | H5 |
| 13:10 | Run 1 at 50% | — |
| 13:50 | Run 2 at 50% | — |
| 14:30 | Runs reviewed; filters sealed | H6 |
| 14:40 | Post-test checks; DG cool-down | H7 |
| 15:15 | Field records reviewed and signed | H8 |
| 15:45 | Close-out meeting: observations, deviations, next steps | — |
Allow spare time. A demonstration that runs out of daylight and has to rush its last runs invites the disputes it was meant to prevent.
What Witnesses Should Actually Watch
Witnesses don’t need to be emission specialists, but they should know what to look for:
- Load: Is kW within tolerance throughout each run? Ask to see the log, not just the controller display.
- Stability: Was the stability criterion met before sampling started?
- Regeneration: Did the controller or differential pressure suggest a regeneration during any run?
- Sampling: Were filters loaded and sealed as described? Were leak checks done and recorded with numbers?
- Isokinetic ratio: Does the lab report each run’s isokinetic ratio, and is it within the method’s limits?
- Deviations: Were any problems recorded on the field sheet, not just discussed?
- Safety: Was access to ports safe throughout?
Encourage witnesses to write down what they see, with times. A witness’s own notes are useful evidence later.
Remote or Partial Witnessing
Sometimes a key witness can’t attend in person, for example a corporate EHS head based in another city or a lender’s engineer. Remote or partial witnessing can work if it is planned rather than improvised.
What can work remotely:
- A live video link at each hold point, with the coordinator showing load logs, leak-check readings and filter seals on camera
- Time-stamped photographs of each filter loading and sealing, port setup and controller displays, shared during the test
- Shared access to live or near-live data, such as load logs and gas analyser readings, through a simple shared spreadsheet or logger export
- Remote sign-off at each hold point, recorded by email or message with the time
What doesn’t work remotely:
- Physical checks of exhaust leaks, port condition and access safety
- Verifying that instruments on site match the serial numbers in the calibration pack, unless someone on site shows each nameplate on camera
- Chain-of-custody signatures on seals, which need someone physically present
A sensible arrangement is at least one buyer-side witness on site, with remote participants joining at hold points. The protocol should list who will attend in person and who remotely, and how remote confirmations will be recorded.
If any party can only witness remotely, agree in advance whether that is acceptable for contractual acceptance. Some contracts and lenders insist on physical presence for payment-linked tests.
Handling Problems on the Day
Problems will happen. What matters is that they are handled according to the protocol and recorded.
| Problem | Typical response |
|---|---|
| Load drifts out of tolerance during a run | Void the run; restabilise; repeat |
| Regeneration starts during a run | Void the run for steady-state purposes; record it; repeat after restabilisation |
| Leak check fails after a run | Void the run; fix the leak; repeat |
| Analyser drift exceeds limit | Flag affected data; recalibrate; repeat if needed |
| Weather makes port access unsafe | Stop; reschedule; never compromise safety |
| Disagreement between witnesses | Coordinator records both positions; testing continues only if the protocol allows |
Every void and repeat should appear in the final report, not just the valid runs. A report that shows only the best runs, with no mention of the others, invites suspicion.
After the Test: From Field Records to Accepted Report
- Field records are signed by all parties before leaving (H8).
- The lab issues a draft report within an agreed time, including the calibration annex, raw data, custody records and all runs (valid and voided).
- Parties review the draft against the protocol, checking calculations, basis and pass criterion.
- Queries are resolved in writing.
- The final report is issued and acceptance is recorded, for example in a signed acceptance certificate.
File the protocol, field records, custody forms, draft and final reports together. They form the evidence trail for the RECD’s performance and will be useful in any later Pollution Control Board inspection, warranty discussion or facility sale.
A Witness Sign-Off Sheet Template
| Item | Confirmed (Y/N) | Comment |
|---|---|---|
| Protocol version used | ||
| Instruments match calibration pack | ||
| Engine and fuel condition as agreed | ||
| Load within tolerance for all valid runs | ||
| Stability criterion met before each run | ||
| No regeneration during valid steady-state runs | ||
| Leak checks and drift checks recorded with values | ||
| Filters sealed and custody forms complete | ||
| Voided runs recorded with reasons | ||
| Safety arrangements followed |
| Name | Organisation | Role | Signature | Date |
|---|---|---|---|---|
| Test coordinator | ||||
| Buyer’s witness | ||||
| Supplier’s representative | ||||
| Laboratory lead |
Frequently Asked Questions
Who should pay for a witnessed RECD test?
That should be set out in the contract. Often the buyer pays for the independent lab, and the supplier attends at its own cost. If a retest is needed because the device fails, the contract should say who pays for that.
Can the RECD supplier’s own lab perform the witnessed test?
It can, but an independent, accredited laboratory gives the result much more credibility, especially if payments or guarantees depend on it. If the supplier’s lab is used, the buyer should insist on full witnessing and raw data.
How many runs are enough?
At least two valid runs per load point is a sensible minimum for a witnessed demonstration, and three is better if time allows. Repeat runs show how consistent the result is and reduce the influence of any single outlier.
Do witnesses need technical qualifications?
Not necessarily. They need to understand the protocol and know what to check. A short pre-test briefing by the coordinator, covering hold points and what each witness should watch, is usually enough.
What if the result is just below 70%?
Follow what the protocol says. That is why the pass criterion, rounding rules and treatment of uncertainty must be agreed in advance. Without that, a borderline result usually leads to a dispute.
Plan the Demonstration Before the Device Is Delivered
The best time to agree a witnessed test protocol is when the RECD is ordered, not when it’s installed. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA, and we regularly take part in witnessed acceptance tests. We can help draft the protocol, plan port locations and coordinate with your chosen lab. See our certifications, learn how an RECD works, or contact our team.