Compressed-Air Leak Detection & Loss Management Engineering

COMPRESSED-AIR LEAK DETECTION & LOSS MANAGEMENT ENGINEERING

A leak is not an energy-saving opportunity until it is found, repaired and verified

A credible programme defines the boundary, detects losses, establishes a defensible estimate, prioritises repairs, controls safe isolation, verifies closure and measures whether plant demand actually falls.

Leak management is also a reliability discipline. Persistent leakage can keep an additional compressor online, reduce storage readiness, increase cycling, raise source pressure and mask genuine production growth.

1 — Define the Leakage Boundary

State which compressors, headers, zones, machines and operating periods are included.

2 — Separate Legitimate Off-Production Demand

Dryer purge, instrument air, continuous process users and drains can consume air when production stops. Identify them before calling residual flow leakage.

3 — Physical Leakage vs Artificial Demand

A leaking fitting is unintended escape; excessive blow-off is avoidable use through intended equipment. Keep the mechanisms distinct.

4 — Leakage Is Dynamic

Leak flow changes with pressure, valve position, machine state and deterioration. One survey is not a permanent leakage percentage.

5 — Ultrasonic Surveys

Ultrasonic instruments can locate turbulent gas leakage; access, background noise, settings and technique affect results.

6 — Acoustic Detection Is Not Automatically Quantitative

A strong signal can prioritise a leak, but converting sound to flow requires a supported method and assumptions.

7 — Off-Production Flow

Where safe and operationally possible, residual system flow can quantify non-production demand after legitimate loads are identified.

8 — Load/Unload Timing Estimate

On suitable fixed-speed systems under controlled no-production conditions, load/unload behaviour can support a leakage estimate. Document capacity and control assumptions.

9 — Pressure-Decay Testing

Controlled isolated-volume decay can support analysis, but volume, temperature, absolute pressure, isolation and active consumers matter.

10 — Worked Residual-Demand Example

Illustrative only: off-production flow is 5.0 m3/min and verified legitimate continuous users consume 1.2 m3/min. The unexplained residual is 3.8 m3/min on that reconciled flow basis; it is not automatically proof that all 3.8 is physical leakage.

11 — Worked Repair Verification

Illustrative only: comparable residual demand falls from 3.8 to 2.4 m3/min after repairs. The observed reduction is 1.4 m3/min under those conditions; reconcile pressure and legitimate loads before annualising.

12 — Record Pressure

A survey at 7.5 bar cannot be compared blindly with one at 6.5 bar.

13 — Reconcile Flow Basis

State the basis before combining leak flow with energy data.

14 — Unique Leak IDs

Assign ID, location, component, date, surveyor, severity, operating pressure and status.

15 — Repairable Location Description

Identify asset, branch, component and physical position so maintenance can find the leak without the surveyor.

16 — Component Classification

Classify fittings, hoses, couplings, drains, valves, regulators, cylinders, manifolds and housings to reveal recurring causes.

17 — Prioritisation

Consider estimated loss, repair difficulty, safety/isolation, production access, recurrence and criticality.

18 — Safety Before Speed

Never tighten or dismantle pressurised components merely to close a leak quickly. Follow isolation/depressurisation procedures.

19 — Difficult Access

Elevated, hot, confined or high-pressure leak locations require competent planning.

20 — Temporary Repairs Need Expiry

Any temporary repair must follow site rules and have a controlled permanent-repair plan.

21 — Hose Leakage

Flexing, abrasion, ageing and poor end fittings can create leaks; inspect condition as well as sound.

22 — Couplings Leak and Restrict

A worn coupling can waste air and create pressure loss. Restore sealing and flow performance.

23 — Threaded Joints

Repeated sealant without correcting damaged threads, alignment or vibration can cause recurrence.

24 — Valve Leakage

External stem leakage and internal bypass are different failure modes.

25 — Cylinder Internal Leakage

Cylinder seals can bypass internally without visible external leakage. Diagnose using a safe machine-specific method.

26 — Solenoid / Manifold Leakage

Valve seats, seals and manifolds can leak internally or externally; machine-level demand testing can reveal hidden losses.

27 — Regulator Venting

Distinguish intentional relieving action from abnormal leakage due to contamination or wear.

28 — Drains

Excessive compressed-air discharge is a loss.

31 — Receiver / Relief Leakage

Suspected vessel or relief-device leakage requires competent inspection; never tamper with safety devices.

32 — Underground Leakage

Zone flow, isolation and pressure trends can narrow buried-line leak searches.

33 — Zone Metering

Submetering assigns residual demand to areas and strengthens accountability.

34 — Isolation Testing

Where safe, sequential branch isolation and flow observation can narrow the loss boundary.

35 — Worked Zone-Isolation Logic

Illustrative only: residual demand falls 1.1 m3/min when one workshop branch is isolated. This directs the survey to that branch; it does not prove one single 1.1 m3/min leak.

36 — Survey Operating State

Some leaks appear only when machines are pressurised or actuating; others are easiest to detect during quiet periods.

37 — Noise Can Mask Detection

Steam, bearings, electrical equipment and production noise can interfere with acoustic methods.

38 — Record Survey Coverage

Distinguish inspected areas from inaccessible, isolated or out-of-scope areas.

39 — Transparent Leak-Cost Formula

State estimated leak flow, energy-response method, operating hours, tariff and assumptions.

40 — Marginal Energy Matters

Removing a small leak may reduce VSD power, unloaded time or an entire compressor's operating hours depending on controls.

41 — Leak Repair Can Remove a Compressor

Crossing a sequencing threshold can create a larger energy effect than a simple proportional estimate.

42 — Leak Repair Can Raise Pressure

After repairs, header pressure can rise if controls are unchanged. Recheck setpoints so saved air does not become excess pressure.

43 — Repair Campaign Planning

Group repairs by area, isolation and production access to improve closure rate.

44 — Spares Strategy

Keep approved high-failure couplings, hoses, seals, drains and fittings where justified.

45 — Standardisation

Repeated failure of one component family can justify controlled standardisation.

46 — Root Cause

Common causes include vibration, damage, contamination, misalignment, ageing, poor assembly and unsuitable components.

47 — Vibration-Induced Leakage

Repeated failures near machinery can indicate unsupported piping or excessive vibration.

49 — Corrosion

Pinhole leakage can signal integrity deterioration; address material condition, not only air loss.

50 — Reliability Warning

A new leak can indicate component deterioration. Safety/reliability significance can exceed energy value.

51 — Verify Every Repair

Reinspect acoustically, visually or by another appropriate method after repair. Close the tag only when confirmed resolved.

52 — Verify System-Level Demand

After a campaign, compare residual or zone demand under similar pressure and operating conditions.

53 — Preserve Before/After Pressure

A lower residual flow at lower pressure is not directly comparable to a higher-pressure baseline without reconciliation.

54 — Preserve Production State

Ensure legitimate loads and production state are comparable when verifying savings.

55 — Do Not Annualise Temporary Conditions

A short shutdown test may not represent annual operation. Use a defensible schedule for annualisation.

56 — Persistence Survey

Repairs can fail and new leaks appear. Repeat surveys according to site condition, recurrence and significance.

57 — Leak Return KPI

Track new leaks found versus leaks closed/verified. A growing backlog indicates deterioration is winning.

58 — Open-Leak Backlog KPI

Report unresolved leaks by severity, estimated loss and age—not count alone.

59 — Verified Repair Rate KPI

Track the share of tagged leaks physically repaired and reverified, not merely work orders issued.

60 — Residual Demand KPI

Trend reconciled off-production demand at comparable pressure/state.

61 — Zone Leakage KPI

Where metering supports it, trend minimum or non-production demand by area.

62 — Financial KPI Needs Evidence

Use verified demand reduction, controlled hours, tariff and an explicit energy-response method.

63 — Carbon Avoidance Needs a Controlled Factor

If emissions are reported, use an approved electricity factor for the relevant period and disclose source/boundary.

64 — Survey Frequency Is Site-Specific

Recurrence, condition and economics govern cadence.

65 — Continuous Monitoring Has Limits

Permanent acoustic or flow analytics can help, but placement, noise, calibration and false positives need governance.

66 — AI Detection Depends on Baseline Quality

Analytics can flag abnormal zone demand or acoustic signatures but cannot replace correct boundaries and verified repairs.

67 — Night-Flow Alarms

Where production state is known, rising minimum flow can trigger investigation before a formal survey.

68 — Maintenance Can Introduce Leaks

After relocation, hose replacement, filter service or piping modification, inspect disturbed connections.

69 — Expansion Changes the Baseline

New equipment adds legitimate volume and leak points. Establish a new controlled baseline.

70 — Temporary Hoses

Rental and shutdown hose networks contain many couplings; inspect them during temporary commissioning.

71 — Contractor Connections

Use approved temporary connections and remove abandoned branches after work.

72 — Leak Survey + Pressure Mapping

A starved zone can have leakage and restrictive piping simultaneously.

73 — Leak Repair + Sequencing

Reduced demand can change optimal base/trim combinations.

74 — Leak Repair + Storage

Lower background demand can improve receiver recharge and transient readiness.

75 — Leak Repair + Capacity Planning

Verified reduction can release capacity for expansion. Do not count estimated unrepaired leaks as future capacity.

76 — Safety Overrides Energy

A relief event, damaged hose or corroded pipe is first a safety/reliability issue.

77 — Survey Documentation Is an Asset

Preserve maps, tag IDs, photos, readings, repair history and verification so future surveys do not restart from zero.

78 — Digital Leak Register

Link location, asset, severity, work order, repair, verifier, flow/cost estimate and recurrence history.

79 — Version Estimation Methods

If ultrasonic conversion, specific-energy basis or tariff assumptions change, version the method so historical reports remain interpretable.

80 — Avoid Inflated Savings Claims

Do not multiply every estimated leak by full-load specific energy and assume all kWh disappear. Control response determines marginal savings.

81 — Unrepaired Leaks Are Not Savings

Opportunity identified is not savings achieved.

82 — Closed Work Order Is Not Verification

A repair record alone is not proof that air loss stopped.

83 — Avoid Double Counting Pressure Savings

If pressure is reduced after leak repair, reconcile interacting effects in the M&V method.

86 — Management Review

Report verified closure, residual demand, recurring causes and unresolved high-value leaks—not merely tag count.

87 — Final Engineering Principle

Leak management is a closed loop: define the boundary, detect, quantify carefully, prioritise, isolate safely, repair root cause, verify the individual leak, verify system demand and repeat before deterioration erases the gain.

88 — Premium Leak Register

89 — Leak Programme KPI Matrix

90 — Technical / Energy / Safety Verification Gate

Survey boundary and excluded areas are documented.

Legitimate off-production loads are separated from unexplained residual demand.

Detection and estimation methods are identified.

Operating pressure and airflow basis are recorded.

Every leak has a unique traceable identity.

Safety/isolation requirements are controlled before repair.

Physical leaks are separated from artificial demand and normal dryer purge.

Repair root cause is considered where recurrence is likely.

Every claimed closed leak is verified.

System-level residual demand is rechecked after material campaigns.

Energy calculations use an explicit marginal/system response method.

Annualisation uses defensible operating hours.

Pressure, production state and legitimate loads are reconciled before/after.

Interacting pressure/sequencing savings are not double-counted.

Persistence survey and ownership are defined.

What is a normal compressed-air leakage percentage?

There is no universal percentage that should be assumed for a specific plant. Measure or estimate the defined system, identify legitimate off-production demand and verify repairs.

Can ultrasonic leak detectors measure leak flow?

Some methods or instruments provide estimates, but quantitative accuracy depends on the supported method, pressure, distance, access and conditions. Preserve the estimation basis.

Is dryer purge a compressed-air leak?

Correct adsorption-dryer regeneration purge is engineered demand, not a physical leak. Abnormal purge or valve leakage should be diagnosed separately.

How do I verify a repaired leak?

Reinspect the individual location and, for material campaigns, compare residual or zone demand under comparable pressure and operating conditions.

Can leak repair reduce compressor capacity requirements?

Verified demand reduction can release system capacity. Capacity planning should use confirmed post-repair demand rather than estimated unrepaired leakage.

Why did system pressure rise after leaks were repaired?

With lower demand and unchanged controls, the system can move toward a higher pressure. Recheck compressor sequencing and approved pressure targets.

Are closed work orders enough to claim savings?

No. A work order documents activity; verification demonstrates that the leak was actually eliminated.

How often should leak surveys be performed?

Cadence is site-specific and depends on recurrence, plant condition, operating hours, economic significance and maintenance capability.

Can AI find compressed-air leaks?

Analytics can flag abnormal demand or acoustic patterns when good data exist, but physical verification, correct boundaries and repair confirmation remain necessary.

What is the best first step in a leak programme?

Define the survey boundary and operating states, then establish how findings will be tagged, repaired and verified before collecting a large backlog of observations.