Installed RECD Sensor Calibration: What Should Facility Teams Document?

The Retrofit Emission Control Device on your generator is probably fitted with a few sensors: a differential pressure transmitter across the filter, perhaps an exhaust back-pressure sensor, and temperature sensors before and after the device. These sensors feed the controller, the Building Management System and the alarms that protect the engine. Every decision about cleaning, regeneration and back-pressure depends on them telling the truth.

Sensors drift. Heat, vibration, soot and moisture in a DG room are hard on instruments. A differential pressure transmitter that reads 8 mbar low can hide a filter that’s close to needing cleaning. A thermocouple with a damaged junction can show a normal temperature while the device overheats.

This guide explains what facility teams should document about installed RECD sensors: what to record when they’re installed, how to check and calibrate them on site, how to record results, what to do when a sensor is found out of tolerance, and how to keep it all organised.

It is different from the calibration of laboratory test instruments used during formal emission tests, which we cover in which instrument calibration records should accompany RECD tests. This article is about the sensors that stay on your RECD every day.

Why Installed Sensors Need Their Own Records

• They make continuous decisions. Alarms, regeneration triggers and maintenance planning rely on them hour by hour.
• They live in harsh conditions. Hot exhaust, vibration, soot and condensation all accelerate drift and damage.
• Their errors look like real changes. A sensor drifting upwards looks exactly like a filter loading with soot. See how to separate a sensor fault from an RECD performance problem.
• Records protect decisions. If a regulator, auditor or supplier questions your data, calibration records show the readings were trustworthy.

Part 1: The Sensor Register

Start with a single list of every sensor whose reading is used for monitoring, alarms or control. This sensor register is the foundation for everything else.

Field	Example
Tag / ID	DG2-RECD-DPT-01
Measurement	Differential pressure across RECD
Location	Tapping points 150 mm before and after RECD body; transmitter on wall bracket 2 m away
Sensor type	Differential pressure transmitter, 4–20 mA output
Make / model / serial number	(as fitted)
Range	0–200 mbar
Output and scaling	4 mA = 0 mbar, 20 mA = 200 mbar
Connected to	RECD controller AI-1; BMS point DG2_RECD_DP
Used for	Monitoring; high ΔP warning and critical alarms; low ΔP alarm
Required accuracy (site tolerance)	±2 mbar
Calibration / check interval	Check every 6 months; full calibration annually
Installed date	(date)
Last check / next due	(dates)

Do the same for each temperature sensor, back-pressure sensor and any pressure switches. For guidance on what each sensor type is suited to, see pressure sensors vs pressure switches in RECD monitoring and RECD temperature sensor placement: what should be measured.

Why “used for” matters

A sensor used only for a local trend can tolerate a looser check regime than one that triggers an engine-protection alarm. Recording what each sensor is used for helps you set sensible tolerances and intervals.

Part 2: Commissioning Records

When the RECD is installed, record each sensor’s starting point:

• Manufacturer’s calibration certificate, if supplied, with serial number matching the installed sensor
• Installation check: the sensor is fitted correctly, impulse lines are routed with the right fall, thermocouples are inserted to the right depth, and cables are protected from heat. See RECD cable routing near hot exhaust lines
• Zero check with the DG stopped (pressure sensors should read close to zero)
• Loop check: the value at the controller, BMS and any remote platform matches the sensor output. See RECD Modbus data: units, scaling and register mapping explained
• Running check: readings at the acceptance-test load points, recorded alongside the lab’s data. These become reference values. See why RECD test results may differ from day-to-day site readings

Part 3: Checks vs Calibrations

Two kinds of activity keep installed sensors honest. Both should be documented.

Routine checks (quick, frequent)

• Zero check: with the DG stopped and cool, pressure sensors should read zero within tolerance
• Comparison check: compare the installed reading with a calibrated portable reference, such as a handheld manometer for pressure or a reference thermometer for temperature, at a steady operating condition
• Plausibility check: inlet and outlet temperatures should be similar with no regeneration, and ΔP should rise with load

These take minutes and catch most problems early.

Calibration (thorough, less frequent)

• The sensor is compared with a traceable reference at several points across its range, for example 0%, 25%, 50%, 75% and 100%
• As-found results are recorded before any adjustment
• The sensor is adjusted if needed, then as-left results are recorded
• This can be done in place, using a portable pressure calibrator or a temperature source, or by removing the sensor to a calibration lab and fitting a calibrated spare meanwhile

Calibration labs should be accredited for the relevant parameter, for example by NABL under ISO/IEC 17025. Portable reference instruments used on site need their own valid calibration certificates.

Part 4: What Each Check or Calibration Record Should Contain

Field	Why it matters
Sensor tag and serial number	Proves which sensor was checked
Date and time	Links to data and event logs
Person and organisation	Accountability
Method (zero check / comparison / multi-point calibration; in place / removed)	Shows how much confidence to place in the result
Reference instrument, serial number and calibration due date	Traceability
Operating condition (DG stopped / load in kW / temperature)	Context for comparison checks
As-found readings at each test point	Shows how far the sensor had drifted
Error at each point, and tolerance	Pass/fail judgement
Adjustment made (yes/no; what)	Explains any change in later data
As-left readings	Shows the sensor’s condition going forward
Pass / fail / replaced	Outcome
Follow-up actions	Including impact assessment if as-found failed

Example: a differential pressure transmitter calibration record

The figures are illustrative.

Test point (applied, mbar)	As-found (mbar)	Error	As-left (mbar)	Error	Tolerance
0	2.4	+2.4	0.1	+0.1	±2.0
50	54.8	+4.8	50.3	+0.3	±2.0
100	107.1	+7.1	100.4	+0.4	±2.0
150	159.6	+9.6	150.6	+0.6	±2.0
200	212.0	+12.0	200.5	+0.5	±2.0

As-found: FAIL. The transmitter was reading high, with the error growing with pressure (a span error plus a small zero offset). After adjustment, as-left: PASS.

Part 5: When a Sensor Is Found Out of Tolerance

An as-found failure doesn’t just mean “fix the sensor.” It raises a question: were decisions made on bad data since the last good check?

Work through an impact assessment:

1. When was the last passing check? The sensor may have been out of tolerance any time since then.
2. What was the sensor used for in that period? Alarms, regeneration triggers, maintenance decisions, reports?
3. What was the effect of the error? In the example above, ΔP read about 10% high at normal operating levels (around 50 mbar), so the high ΔP warning would have triggered early, which is a nuisance rather than a risk. If the error had been low, alarms could have triggered late, which is a real risk.
4. Were any reported figures affected? For example, ΔP trends used to plan cleaning, or monthly KPIs. See what should a useful RECD operating dashboard show.
5. Is any correction needed? For trends, you may annotate the period or apply a documented correction. Never overwrite original data.
6. Should the interval be shortened? Repeated failures suggest the check interval is too long, or that the sensor or its installation needs improving.

Record the assessment in the event log. See what should an RECD event log record.

Part 6: Setting Tolerances and Intervals

Tolerances

Set the site tolerance for each sensor from:

• What it’s used for. An engine back-pressure alarm needs a tighter tolerance than a trend-only reading.
• The distance between normal readings and alarm limits. If normal ΔP is 45 mbar and the critical alarm is 80 mbar, a ±2 mbar tolerance is comfortable. If normal back-pressure is 52 mbar and the limit is 60 mbar, ±2 mbar is tight and may need a better sensor.
• The sensor’s specified accuracy. Don’t set a tolerance tighter than the sensor can achieve.

Intervals

Start with the supplier’s recommendation, then adjust based on what you find:

As-found history	Action
Passes comfortably several times in a row	Consider extending the interval
Passes, but near the tolerance limit	Keep the interval
Fails	Shorten the interval; investigate cause
Fails repeatedly	Review the sensor type, installation and environment

Recording as-found values over time is what makes this evidence-based. Without as-found data, you can’t tell whether your interval is right.

Part 7: Practical Issues Specific to RECD Sensors

Pressure impulse lines

Differential pressure and back-pressure sensors are usually connected to the exhaust through small tubes (impulse lines). These can:

• Block with soot, causing readings to freeze or respond slowly
• Fill with condensate if not routed with a continuous fall back to the exhaust, causing offset readings
• Leak at fittings, causing low readings

A calibration of the transmitter alone won’t reveal impulse-line problems. Include an impulse-line inspection and blow-through, following the supplier’s instructions, in the check routine, and record it.

Thermocouples and RTDs

• Thermocouples can degrade at high temperature, and their junctions can fail. Check them against a reference at operating temperature if possible, or by comparing inlet and outlet readings when no regeneration is active.
• Extension cables must be the correct type for the thermocouple. Wrong or damaged cables introduce errors.
• RTDs (such as Pt100) are stable but can be damaged by vibration. Check wiring and connections.

The NIST temperature scale and thermocouple reference database is a useful technical reference for thermocouple types and their standard output tables.

Controller and BMS scaling

A sensor can be perfectly calibrated but still show wrong values if the controller or BMS scaling has changed, for example after a firmware update or controller replacement. Include a loop check in each calibration: apply a known input and confirm the value at the controller, BMS and any remote platform.

Replacement sensors

When a sensor is replaced, record:

• New serial number and certificate
• Any change in range or output type
• Updated scaling in controller and BMS
• Zero and loop checks after installation
• An event log entry, so later readers know why the trend may have shifted slightly

Part 8: Who Owns What

Calibration slips through the cracks when everyone assumes someone else is doing it. Agree responsibilities:

Task	Facility team	RECD supplier	DG maintainer / calibration provider
Maintain the sensor register	Owns	Provides initial data	Updates after work
Set tolerances and intervals	Approves	Recommends	Advises
Routine zero and comparison checks	Arranges	May perform under contract	May perform
Full calibration	Arranges	May perform under contract	Performs with traceable references
Impulse line inspection	Arranges	Specifies method	Performs
Loop checks after controller or BMS changes	Arranges	Supports	BMS contractor performs
Impact assessment after failures	Owns	Advises	Supports
Record keeping and filing	Owns	Supplies certificates	Supplies reports

The facility team doesn’t need to do the technical work, but it must own the register and the records. Otherwise nobody notices when checks fall overdue.

Part 9: A Worked Example of Undocumented Drift

The scenario is illustrative. A hospital’s DG set had an engine back-pressure transmitter feeding a high back-pressure alarm set at 55 mbar, just under the engine maker’s limit of 60 mbar. The transmitter had been installed with the RECD three years earlier, and nobody had checked it since.

During a long summer outage, the turbocharger showed signs of overheating. The back-pressure alarm had never triggered.

A calibration check found the transmitter reading about 9 mbar low across its range. When the BMS showed 51 mbar, the true back-pressure was around 60 mbar, right at the engine’s limit. The cause was a combination of zero drift and a partly blocked impulse line.

Because there were no as-found records, nobody could say when the drift began, so it wasn’t possible to tell how many earlier runs had exceeded the limit unnoticed.

The site then:

• Added the transmitter to a sensor register with a six-monthly check and an annual calibration
• Added impulse line inspection to the same schedule
• Added an independent back-pressure switch as a hardwired backup alarm. See pressure sensors vs pressure switches in RECD monitoring
• Recorded as-found results at each check, so future drift could be detected and its start dated

A single as-found record a year earlier would probably have caught the drift before it mattered.

Part 10: Labelling in the Field

A simple label on each sensor helps technicians and auditors:

• Tag number
• Last calibration or check date
• Next due date
• Initials of the person who did it

Labels don’t replace records, but they make overdue sensors obvious during walk-rounds.

A Calibration Record Template You Can Adapt

Field	Entry
Sensor tag / serial
Measurement and range
Date / time / performed by
Method (zero / comparison / multi-point; in place / removed)
Reference instrument, serial, cal due
Operating condition
Test points and as-found readings
Errors and tolerance
Adjustment made
As-left readings
Impulse line / wiring inspection result
Loop check (controller / BMS / platform values)
Result: pass / fail / replaced
Impact assessment needed? Reference
Next due date
Signature

Where These Records Fit

Sensor calibration records belong with the RECD’s other records: certification, test reports, maintenance history and event logs. Together they form the evidence that the device is monitored and maintained properly. See the RECD maintenance guide and the environmental compliance checklist for generator owners. They’re also worth having ready for a Pollution Control Board inspection.

If calibration is part of a supplier service contract, the contract should say which sensors are covered, at what intervals, and what records will be handed over. See RECD service response commitments.

Related Reading

• Pressure sensors vs pressure switches in RECD monitoring
• How to compare RECD pressure readings at different generator loads
• RECD temperature sensor placement: what should be measured?
• How to separate a sensor fault from an RECD performance problem

Frequently Asked Questions

How often should RECD sensors be calibrated?

Follow the supplier’s recommendation to start with, often a quick check every few months and a full calibration annually. Then adjust based on as-found results. Sensors that drift often need shorter intervals, and stable ones may justify longer intervals.

Can we calibrate RECD sensors without removing them?

Often, yes. Pressure transmitters can usually be calibrated in place with a portable pressure calibrator, by isolating them from the exhaust. Temperature sensors can be checked by comparison with a reference probe, or removed and checked in a portable temperature source. Follow the supplier’s instructions.

What is the difference between as-found and as-left?

As-found is the reading before any adjustment, showing how much the sensor had drifted since the last calibration. As-left is the reading after adjustment, showing its condition going forward. Both should be recorded. As-found data is essential for setting intervals and assessing the impact of drift.

Do we need a lab certificate for every installed sensor?

The sensor’s initial certificate and the certificates of the reference instruments used for on-site checks are what matter most. Routine checks done with a calibrated reference are acceptable if they’re well documented.

What should we do with data recorded while a sensor was out of tolerance?

Keep it, but don’t treat it as reliable. Annotate the affected period in the data and event log, note the size and direction of the error, and assess whether any decisions or reports relied on it. If a correction is applied for analysis, store it separately and document how it was calculated. Never overwrite the original readings.

Do pressure switches need calibration records too?

Yes. Record each switch’s as-found and as-left trip points at every test, together with the impulse line condition. A switch’s setpoint can drift without any visible sign, so test records are the only proof it still trips where it should.

Who should do the calibration?

A competent technician from the RECD supplier, the DG maintainer or a calibration service provider, using calibrated reference instruments. The facility team should own the register and the records, whoever does the work.

Trust Your Readings, and Be Able to Prove It

Sensors are the RECD’s eyes. Keeping them calibrated and documented is what makes every alarm, trend and report trustworthy. Aceget supplies CPCB type-approved Retrofit Emission Control Devices for DG sets from 25 kVA to 2500 kVA, and we can provide sensor registers, check procedures and calibration support for our installations. See our certifications, or contact our team.