- October 12, 2026
- Posted by: Aceget
- Category: RECD & Emission Control Devices
An RECD trial installation is a sensible first step when you own many diesel generator sets and are not yet sure how a retrofit emission control device will behave on all of them. Campus operators, telecom circles, mall groups, hospital chains and multi-plant manufacturers often ask for one set to be fitted first, run for a while, and judged before the remaining sets are ordered. The same request comes from single-site owners whose duty pattern is uncertain, for example a standby set that runs a few minutes a week and occasionally carries a heavy load for hours. The trouble starts when nobody writes down, before fitting, what success looks like. A trial without agreed criteria tends to end in an argument rather than a decision.
This guide treats the trial as a small engineering experiment: pick the right set, measure a baseline, define what will be judged and how, collect data with discipline, and agree in advance what each result means commercially.
Trial, Pilot or Commissioning? Getting the Terms Straight
People use these three words loosely, and the confusion causes real disputes. Agree the term at the start and write its meaning into the order.
| Aspect | Commissioning / SAT | Trial installation | Pilot rollout |
|---|---|---|---|
| Purpose | Confirm a purchased unit is installed correctly and works on day one | Find out whether the device suits this type of set and duty before committing to more | Prove installation method, logistics and service model on a small group of sets |
| Number of sets | Every set in the order | Usually one or two representative sets | A batch, often one site or one region |
| Typical duration | Hours to a day | A defined number of running hours, often spread over weeks | Weeks to a few months |
| Who pays | Included in the supply or turnkey price | Negotiated: paid, part-paid or returnable, with removal terms agreed | Normally a firm order at agreed rates |
| Outcome | Signed site acceptance record | Go/no-go decision on wider use | Lessons applied to the full rollout schedule |
Commissioning checks are covered in detail in our piece on factory acceptance versus site acceptance tests for RECD projects.
One point needs to be clear from the outset. A trial does not certify anything. The particulate capture efficiency of an RECD model comes from type approval testing under the procedure in the Central Pollution Control Board circular dated 01.02.2022 on the system and procedure for emission compliance testing of RECDs for DG engines up to 800 kW gross. That testing is carried out by recognised agencies such as ARAI, Pune and ICAT, Manesar. A field trial on your site checks fit, engine behaviour, maintenance practicality and how the device copes with your duty. It cannot replace certification, and if the set will be operated for compliance purposes during the trial, the unit fitted must be a certified model matched to the engine. Before agreeing anything, check that the RECD is CPCB-tested for your DG set.
Choose the Right Set for Your RECD Trial Installation
The set you choose decides how much the trial tells you. A perfect result on the newest, best-loaded set in the fleet proves very little about the twenty-year-old unit in a cramped basement. Look at three things when shortlisting.
- Load profile. Pick a set whose loading looks like most of the fleet. If you have not studied this yet, start by using DG load logs to assess RECD suitability. Monthly running hours, average load and the share of time spent at light load all matter.
- Exhaust layout. Long runs, several bends, a rooftop discharge or a tight canopy all add resistance and limit access. A set with a difficult layout tests the installation design properly.
- Engine age and condition. Choose an engine of typical age and service history. An engine that is overdue for overhaul will produce a result that says more about the engine than the device.
Remember too that two sets with the same kVA badge can carry different engines and need different devices, as explained in why same-kVA DG sets may need different RECDs.
Representative vs worst case
There are two honest strategies. A representative trial picks the most typical set and tells you what will happen on most of the fleet. A worst-case trial picks the hardest set, perhaps the lightly loaded standby unit with frequent short starts, and tells you whether the device survives the toughest duty you have. Light loading is a known challenge, covered in choosing an RECD for a generator running at low load, and so are repeated brief runs, discussed in RECD selection for DG sets with frequent short start-stop cycles.
If budget allows, trial one of each. If not, say clearly in the trial document which approach you have chosen. A worst-case trial that needs one extra cleaning should not be judged as a failure if the criteria were written for a typical set, and a representative trial should not be used to promise performance on the oddest set in the fleet.
Record a Baseline Before Anything Is Fitted
Most trial disputes come down to one sentence: “It was not like this before.” Without a baseline, nobody can prove or disprove it. Spend a day measuring the set as it is, with the supplier’s engineer and your own maintenance team present.
- Engine condition. Note running hours, last service date, visible smoke at start and under load, and any known faults. Record lube oil top-up history; high consumption is a warning sign, as described in how engine oil consumption affects RECD suitability.
- Existing exhaust backpressure. Measure at a fixed tapping point at the highest load you can safely apply. The silencer and pipework already use part of the engine’s allowance, so this number shows how much headroom is left. Our explainer on how exhaust backpressure affects a DG set sets out why this matters.
- Exhaust temperature at load steps. Record temperature at, say, light, medium and the highest practical load, holding each step long enough to settle.
- Noise at fixed points. Mark measuring points with paint or tape and record readings with the same meter at the same load.
- Load profile. Extract several months of logs: hours per month, number of starts, typical load and peak load.
- Photographs. Take dated photos of the exhaust route, supports, canopy and stack outlet.
If the baseline shows black smoke or poor combustion, sort that out first. The causes are covered in why a generator emits black smoke. Fitting a particulate device to an engine in poor health only shortens the cleaning interval and muddies the result.
Writing Criteria That Can Actually Be Judged
“The DG should run normally” is not a criterion. Neither is “backpressure should be acceptable”. Each criterion in the trial document should answer seven questions:
- Parameter: what exactly is being judged.
- Method: how it will be measured or checked.
- Instrument: which gauge, meter or logger, and its calibration status.
- Point: where on the set or exhaust the reading is taken.
- Load condition: the load at which the reading counts.
- Limit: the pass value, target or action threshold, and where that value comes from.
- Recorder: who takes the reading and who countersigns it.
Where a limit belongs to someone else, cite them. Backpressure limits belong to the engine OEM, not the RECD supplier and not the buyer. Capture efficiency belongs to the type approval certificate. If you are writing a tender later, the same discipline applies; see writing RECD tender specifications with measurable performance requirements. Not every criterion carries the same weight, so sort them into three classes.
Must-pass (gate) criteria
These decide the outcome on their own. If one fails and cannot be corrected within the trial, the answer is no. Keep this list short: certified model matched to the engine, total backpressure within the engine OEM’s permitted limit, no exhaust leaks, and no abnormal engine behaviour attributable to the device. A long gate list invites failure on trivia.
Monitored criteria with action thresholds
These are tracked throughout and trigger a defined response when a threshold is crossed. Backpressure rise over the trial is the obvious one. Rather than “fail if backpressure rises”, write “if backpressure reaches the agreed alert level, the supplier carries out cleaning and the interval is recorded”. The trial then measures the real cleaning interval on your duty, which is exactly what you need to plan maintenance for the fleet.
Observations recorded for information only
Some things are worth noting but should not decide the result: the appearance of the stack outlet, small noise changes at points nobody occupies, operator comments, or ease of access for a particular task. Record them, discuss them at the review, and use them to improve the rollout design.
A Sample Acceptance Criteria Table
The table below is a starting point, not a standard. Adjust classes and limits to your engine, site and contract. Wherever it says “OEM limit” or “declared value”, fill in the actual figure from the engine manufacturer or the supplier’s written offer before the trial starts.
| Criterion | Class | Method | Limit / target | When measured |
|---|---|---|---|---|
| RECD model and certification match engine gross kW | Must-pass | Compare certificate, nameplate and engine data plate | Model certified for the engine’s gross kW range | Before installation |
| No exhaust leaks at joints, flanges or bellows | Must-pass | Visual and soot-mark check, joints inspected under load | No visible leakage or soot tracking | Commissioning and each inspection |
| Total backpressure at highest practical load | Must-pass | Calibrated gauge or manometer at fixed tapping point | Within engine OEM’s permitted limit | Commissioning, then at each scheduled reading |
| Backpressure rise over trial and cleaning interval | Monitored | Trend of readings at same load and point | Alert level and cleaning interval as declared in supplier’s offer | Every logging interval |
| Exhaust temperature behaviour | Monitored | Thermocouple or infrared reading at agreed points, compared with baseline | No unexplained change from baseline at same load | Each load step test |
| No alarm or controller faults | Must-pass | Review RECD and DG controller alarm history | No unresolved faults caused by the device | Weekly and at review |
| No abnormal engine parameters | Must-pass | Coolant temperature, power derate, black smoke under load | No change from baseline attributable to RECD | Each load run |
| Noise change at agreed points | Monitored | Same meter, same points, same load as baseline | No increase beyond the agreed margin | Commissioning and end of trial |
| Vibration and supports | Monitored | Visual check of supports, bolts, bellows and hangers | No cracking, loosening or pipe movement beyond design | Each inspection |
| Visual smoke at stack | Observation | Photograph at start and under steady load | Recorded for comparison with baseline | Start and end of trial |
| Maintenance tasks feasible within service clearance | Must-pass | Supplier demonstrates removal or cleaning access on site | Tasks completed without dismantling unrelated equipment | During trial, ideally at first cleaning |
| Documentation complete | Must-pass | Check certificate copy, O&M manual, as-built sketch, warranty | All items handed over | End of trial |
| Site emission test by NABL lab | Optional | Stack test by an accredited laboratory | As required by consent conditions | Only if consent or regulator requires |
A few rows deserve comment. On noise, agree the margin in advance and note that ambient noise at a busy site can swamp small differences. On vibration, the principles in how to prevent generator vibration issues apply equally to the added weight of an RECD on the exhaust line. On maintenance access, the device should be serviceable within the space planned for it, as described in service clearance around an RECD. On the optional emission test, a smokeless stack is not proof of particulate capture, which is why a smokeless DG set may still need particulate measurement. If your consent conditions call for a stack test, use a laboratory accredited by NABL and agree who books and pays for it.
How Long Should the Trial Run?
Define the RECD trial period in engine running hours, not calendar days. A prime set running most of the day collects meaningful data quickly. A standby set may log only a few hours in a month, so a “60-day trial” could end with almost no evidence. Agree a minimum number of running hours, a minimum number of starts, and a calendar cap so the trial cannot drag on indefinitely.
Standby sets usually need help. Schedule load runs on the set’s own building load where safe, or hire a load bank for defined sessions. The difference between duty types is explained in standby versus prime power generators.
| Duty type | What to agree | Practical notes |
|---|---|---|
| Prime or continuous | Running hours covering at least one expected cleaning cycle, plus a calendar cap | Natural operation usually provides enough data; record load at each reading |
| Standby with regular outages | Minimum running hours and number of starts, with planned load runs to fill gaps | Coordinate with outage season; monsoon and summer often bring more hours |
| Standby with rare outages | Scheduled load bank sessions at defined loads, plus weekly test runs logged | Factor load bank hire and fuel into the trial cost |
| Peak shaving or cyclic | Hours and start count, with a review after a set number of cycles | Short cycles are the key variable; log every start |
| Telecom or unattended sites | Hours from remote monitoring, site visits at fixed intervals | Agree who visits and how readings reach the review team |
Data Collection Plan
A trial is only as good as its log sheet. Decide before installation who records what, how often, and where the sheets are kept. Joint readings by the buyer’s technician and the supplier’s engineer carry far more weight at the review than either party’s numbers alone.
| Who | What | Frequency |
|---|---|---|
| Site DG operator | Running hours, starts, load, alarms, visible smoke | Every run |
| Maintenance technician | Backpressure and exhaust temperature at agreed load, leak and support check | Weekly or per agreed hours |
| Supplier engineer with site technician | Joint readings, cleaning if triggered, photographs | At set intervals and at each cleaning |
| Facility or project manager | Review of sheets, escalation of deviations | Fortnightly |
Keep the log sheet simple enough for a night-shift operator. Useful fields:
- Date, time, hour meter reading and cumulative trial hours
- Load in kW or percentage, and whether it was building load or load bank
- Backpressure reading, instrument ID and tapping point
- Exhaust temperature at the agreed points
- Coolant temperature and any derate or alarm
- Smoke observation at start and under load
- Remarks, initials of recorder and countersignature
Photographs should be dated and taken from the same positions each time: the gauge face, the joints, the supports and the stack.
Making the Decision
The go/no-go decision should not be a single yes or no. Four outcomes cover nearly every real trial, and each should have a commercial consequence written into the trial agreement before the device is installed.
| Outcome | When it applies | Commercial consequence |
|---|---|---|
| Accept | All must-pass criteria met; monitored criteria within thresholds | Trial unit purchased at the agreed price; rollout order released at agreed rates |
| Accept with conditions | Must-pass criteria met; a monitored item needs a design change or maintenance plan | Trial unit purchased; rollout order includes the agreed change, such as a revised layout or cleaning schedule |
| Extend | Not enough running hours or a disputed reading | Trial extended by a fixed number of hours, once only, at no extra charge unless agreed otherwise |
| Reject | A must-pass criterion failed and could not be corrected | Device removed and original exhaust restored at the cost agreed in advance; any deposit refunded or credited as agreed |
Before calling a rejection, check the cause. If backpressure rose quickly because the engine was burning oil, or the set ran far outside the conditions the supplier declared, the result may say more about the operating conditions than the device. Our article on RECD performance guarantees and operating conditions explains why declared conditions matter. Hold the review meeting with the site team, the DG service provider and the supplier together, go through the sheets line by line, and sign a short decision record.
Commercial Terms of a Trial
Engineering criteria are only half the agreement. The other half is money and time. Settle these points in writing:
- Price of the trial unit. Whether it is paid in full, part-paid with balance on acceptance, or supplied against a refundable deposit.
- Installation and removal. Who pays to fit it, and who pays to remove it and restore the original exhaust if the result is a rejection.
- Conversion to order. How the trial unit becomes a purchased unit, and how any deposit is adjusted.
- Rollout pricing. Rates for the remaining sets, how long they are held, and how quantities affect them.
- Exclusivity. Whether the buyer is trialling other suppliers in parallel, and whether that is acceptable to both sides.
- Timeline cap. A latest date for the decision, so neither party is left waiting.
- Changes. How design changes arising from the trial are handled; the same logic as in managing scope changes after an RECD order.
If you plan to visit the supplier’s works before or during the trial, prepare using questions to ask during an RECD factory visit.
Worked Example
The following example is hypothetical and illustrative. It does not describe a real customer, and the outcomes shown are qualitative.
A hospital group runs twelve DG sets across four hospitals. Most are standby units that run during grid outages and weekly tests, but two sites have unreliable supply and run their sets for long periods. The engineering head wants an RECD pilot installation before committing to all twelve, and the clinical side insists that nothing must compromise backup power. Guidance on this sector appears in RECDs for hospitals and critical backup loads.
The group picks two sets. Set A is a lightly loaded standby unit with frequent short test runs, the worst case for particulate loading. Set B is a prime-duty set at the site with poor grid supply, representative of the two heavy-use sites. Baselines are recorded for both. For Set A, the trial agreement includes monthly load bank sessions because natural running hours are too low. The criteria use the table above, with the backpressure limit taken from the engine OEM’s documentation and the alert level and cleaning interval taken from the supplier’s offer.
| Criterion | Set A (standby, light load) | Set B (prime duty) |
|---|---|---|
| Certification match | Pass | Pass |
| Exhaust leaks | One flange weep found at first inspection, corrected | Pass |
| Backpressure within OEM limit | Pass | Pass |
| Backpressure rise and cleaning | Reached alert level sooner than declared interval | Within declared interval |
| Engine parameters | No change from baseline | No change from baseline |
| Maintenance access | Pass after relocating a cable tray | Pass |
| Decision | Accept with conditions | Accept |
For Set A, the condition was a revised maintenance plan: a regular loaded run to support regeneration and a shorter inspection interval for similar standby sets. The rollout order for the other ten sets grouped them into “standby, light load” and “prime duty” families, each with its own maintenance schedule. A telecom operator could follow the same approach, with remote logging replacing site operators; see RECDs for telecom sites with remote unattended operation. Data centres, where test runs are frequent and loads are carefully controlled, are covered in RECDs for data centres and redundancy testing, and mall groups coordinating several tenants can read coordinating DG retrofits across mall tenants.
Mistakes That Make Trials Inconclusive
- Starting the trial with no baseline readings, then arguing about whether a change was caused by the device.
- Defining the trial in calendar days on a standby set that barely runs.
- Taking backpressure readings at different loads, different points or with different instruments, so the trend means nothing.
- Using a backpressure limit invented by the buyer or supplier instead of the engine OEM’s figure.
- Fitting an uncertified or unmatched model “just for the trial” and then running the set for compliance.
- Choosing the best set in the fleet and assuming the result applies to the worst.
- Ignoring an engine fault, such as heavy oil consumption, that loads the device faster than normal.
- Leaving the commercial outcome of a rejection undefined, so removal becomes a dispute.
- Changing the trial set’s duty during the trial without recording it.
- Having no single named person on each side responsible for the log sheets.
Once the trial is accepted, routine care continues as described in our RECD maintenance guide, using the cleaning interval actually measured during the trial.
Frequently Asked Questions
Does a successful trial prove the RECD meets the capture efficiency requirement?
No. Capture efficiency is established through type approval testing under the CPCB procedure dated 01.02.2022. In the NCR, CAQM directions refer to RECDs with at least 70% particulate capture for in-use DG sets up to 800 kW. A field trial checks how the certified model behaves on your engine and duty. To confirm what was tested, learn how to read an RECD test certificate.
Who sets the backpressure acceptance limit?
The engine OEM. The trial should use the maximum exhaust backpressure published by the engine manufacturer for that engine and rating. The RECD supplier declares the device’s own resistance and the expected rise between cleanings, but the total system figure must stay within the engine OEM’s permitted limit.
Can we run the trial on a set that only runs for weekly tests?
Yes, but plan for it. Add scheduled load runs or load bank sessions so that the set accumulates enough running hours at meaningful load. Otherwise the trial ends with little data.
Should one trial cover every set in our fleet?
Only if the sets are similar in engine model, rating, duty and exhaust layout. Mixed fleets usually need one trial per group, or at least one representative set and one worst-case set.
What happens to the trial unit if we reject it?
That depends on what was agreed before installation. Typical terms cover who removes the device, who restores the original exhaust, who pays for that work, and how any deposit or advance is refunded or credited. If none of this was written down, expect a negotiation.
Conclusion
An RECD trial installation is worth doing when you have many sets or an uncertain duty pattern, but only if it is set up to produce a decision. Choose the set deliberately, record a baseline, write criteria that name the parameter, method, point, load and limit, and sort them into must-pass, monitored and observation classes. Measure the trial in running hours, keep joint log sheets, and agree before installation what Accept, Accept with conditions, Extend and Reject mean in money and time.
If you are planning a pilot before rollout across multiple DG sets, take a look at our CPCB-certified RECD range and talk to our team about trial terms, baseline readings and acceptance criteria suited to your sets.