Interactive resource
Use the working tool for this page
This page has a matching worksheet, calculator or comparison resource. It runs in your browser: enter your own measured values and keep the evidence behind every result.
Open resourceCompressor PerformanceAcceptance Worksheet
Confirm the delivered duty, measurement boundary and evidence—before calling a compressor accepted.
1. Engineering answer: acceptance needs more than a running machine
A compressor performance acceptance worksheet connects an agreed requirement to a valid test, a traceable measurement and a recorded decision. It establishes whether the identified equipment delivers the required airflow at the specified pressure, with the agreed electrical input and any separately required quality, control or condition performance. A successful start, a steady pressure gauge or a completed installation checklist does not answer all of those questions.
The practical sequence is to define the requirement, confirm the test conditions, measure the correct quantities, apply only authorised corrections and evaluate the result against the decision rule agreed before testing. Keep a numerical comparison separate from the final acceptance decision: missing evidence, an unsuitable test state or an unresolved critical defect can prevent release even when one displayed number appears satisfactory.
It supports new equipment, replacement packages and post-overhaul checks without treating them as identical assignments. The relevant configuration and the scope of the promised performance must remain visible throughout. The objective is a result another engineer can reproduce—not a favourable percentage produced by changing the comparison after the measurements are known.
2. The decision this worksheet owns
This page owns the acceptance calculation and its supporting evidence record. It does not reproduce an entire commissioning manual, select equipment from scratch or provide contractual remedies.
A practical result has three layers. Test validity establishes whether the observations can answer the specified question. Numerical evaluation compares valid results with the frozen requirement. Release disposition records what the responsible authority accepts, with any remaining actions. Combining these into a single green indicator makes it difficult to distinguish a failed machine from an invalid measurement or an incomplete handover.
Use the worksheet for a defined compressor package or an explicitly wider compressed-air system. Where a project requires a formal test code, record its exact identity, edition and applicable provisions in the project file. This worksheet does not supply missing code clauses, manufacturer corrections, contractual tolerances or site-specific acceptance limits. A missing requirement remains an open item rather than becoming a generic website default.
3. Freeze a requirement that can actually be tested
Copy each acceptance requirement from the accepted specification, agreed supplier document or approved change. Assign a requirement ID and preserve its source revision. Record the quantity, limit direction, unit, measurement location, operating state and reference basis. “Capacity satisfactory” is not a measurable requirement; neither is “as per catalogue” when several catalogue issues or pressure ratings exist.
Separate a nominal rating from the actual acceptance limit. A stated allowance may define a lower airflow limit or an upper power limit, but its applicability must be explicit. Do not add a second allowance after the contract already incorporates one. Record whether the criterion applies to a stable mean, each individual run, an instantaneous minimum, an interval maximum or a complete operating cycle.
Requirement | Definition needed before testing | Evidence needed for comparison |
|---|---|---|
Delivered airflow | Minimum value, pressure, reference conditions, boundary | Identified flow method and accepted test interval |
Package power | Maximum input, included auxiliaries, load state | True electrical input on the agreed feeder boundary |
Specific power | Maximum ratio, aligned airflow and power basis | Simultaneous or properly synchronised measurements |
Critical-user pressure | Minimum, maximum or permitted excursion rule | Correct location, sampling interval and duty event |
Air quality or dew point | Parameter, unit, sampling point and operating conditions | Appropriate sampling and measurement evidence |
Controls and condition | Required response, temperature or vibration criterion | Defined procedure, state, result and applicable limit |
A requirement without an agreed limit can still support descriptive reporting. It cannot support an invented pass/fail decision. Preserve the unresolved question and the person responsible for resolving it.
4. Identify the equipment and the boundary being accepted
Record the compressor tag, make, model, serial number, drive arrangement, control mode and relevant software or settings revision. For rebuilt equipment, connect the installed airend identity and repair scope to the package. A report for another serial number or an earlier configuration can support background comparison, but it is not automatically evidence for the delivered unit.
Define the physical inlet and outlet, power supply point and included accessories. A package outlet upstream of a dryer is a different air boundary from a treated-air header downstream of purge, drains and distribution losses. Main-motor input is a different power boundary from the complete compressor package or the compressor-room system.
Document connections capable of influencing the result: other compressors, a shared receiver, bypasses, temporary lines and externally supplied air. Preserve the actual valve and control configuration through a drawing reference or clear connection description. The record should explain where every measured flow can originate and where it can leave. An accurate meter cannot make an ambiguous boundary valid.
5. Select test points around the promised operating duty
Define the full-load point and any required part-load, unloaded, minimum-speed or transition conditions before starting. Test points should follow the accepted duty and equipment capability, not an arbitrary number of convenient percentages. A plant buying a trim compressor may need evidence across its supported operating range, while a base-load duty may emphasise a different set of conditions.
For each point, record the target pressure, target flow or speed, permitted operating envelope, stabilisation criteria, observation period and repeat-run rule. Specify what counts as an interrupted run and which conditions require a new run. The required duration should be technically justified for the equipment and variable being assessed; there is no universal five-minute or one-hour acceptance window in this worksheet.
Retain chronology. A load transition, dryer regeneration event or cooling-fan cycle may be relevant rather than disposable noise. Identify whether it belongs in the steady-state comparison, the cycle-energy comparison or a separate functional test. A favourable steady segment should not replace a required complete-cycle observation.
Agree witness responsibilities and record attendance against the specific test. A witness signature establishes what was observed within its stated scope. It does not silently certify instrument accuracy, a proprietary correction method or every discipline involved in the installation.
6. Distinguish published data, factory evidence and site evidence
Published performance data help establish a comparison basis; factory acceptance testing and installed site testing answer different questions. A factory result cannot by itself demonstrate the customer's pipework, cooling-air arrangement, electrical supply, treatment train or critical-user pressure. Equally, an unsuitable site test should not be represented as a valid rejection of a correctly defined factory performance claim.
CAGI describes its performance-verification programme as independent testing of participating manufacturers' published performance information using sampled equipment. That is useful evidence for the applicable model and published conditions, not a substitute for identifying the supplied unit and completing its agreed site acceptance. Check the relevant sheet and programme scope rather than assuming that every data sheet or manufacturer logo denotes verified performance. [R1]
For each evidence item, state whether it is a published rating, model-level verification, factory test of the identified unit, field measurement, calculation or unverified supplier statement. Preserve the file, revision, conditions and relationship to the requirement. The distinction makes it possible to accept a documented factory requirement while leaving a separate installed-system requirement open.
7. Confirm readiness and measurement capability
Performance testing begins after the responsible personnel establish safe readiness. Appropriate completion, isolation arrangements, pressure-rated test connections, guards, utilities and controlled discharge arrangements are prerequisites—not acceptance shortcuts. Do not improvise a restriction, defeat a protective function or dismantle a pressurised component to obtain a convenient test point. The actual test procedure belongs to competent project personnel.
The instrument register should identify what each device measures, where it is installed, its range, relevant configuration and calibration or verification evidence. Check that the method is suitable at the expected reading and under the actual flow, pressure, temperature and electrical conditions. A calibration certificate does not resolve an unsuitable installation.
Instrument or channel | Record in the worksheet | Common interpretation risk |
|---|---|---|
Airflow meter | Principle, tag, location, range, output unit, reference basis | Treating a normalised output as actual compressed volume |
Pressure sensor | Tap location, gauge or absolute basis, range | Using header pressure as the pressure at the flow meter |
Temperature sensor | Location, immersion or contact arrangement, unit | Substituting room temperature for the measured gas temperature |
Power analyser or energy meter | Supply boundary, wiring/configuration, time interval | Omitting separately supplied auxiliaries |
Quality or dew-point instrument | Sample point, pressure, conditioning and method | Comparing unlike sample conditions |
Data logger | Clock, interval, averaging and missing-data behaviour | Pairing readings from different operating states |
Record instrument substitutions as changes with effective times. Do not quietly join before-and-after records when scaling or measurement basis changed.
8. Make airflow meaning unambiguous
The worksheet must distinguish actual volume at the meter from a volume expressed at specified reference conditions. A label such as cfm, normal cubic metres or free-air delivery is insufficient unless the associated pressure, temperature, humidity convention and test basis are available. Preserve the original value and unit alongside the accepted comparison value.
A geometric conversion changes only the unit: one cubic foot is 0.028316846592 cubic metres. It does not convert actual compressed volume into free-air delivery, change a dry basis to a wet basis or reconcile two different reference temperatures. The reference conversion and the unit conversion should remain separate recorded operations.
For an illustrative dry-gas conversion with unchanged composition, an equation derived from conservation of molar flow is:
Q_ref = Q_m × (p_m,abs / p_ref,abs) × (T_ref / T_m) × (Z_ref / Z_m)
Here Q_m is actual volumetric flow at the meter, T is absolute temperature in kelvin and Z is compressibility factor. The equation transfers a gas quantity between stated volumetric bases; it does not predict how the compressor itself would perform at a different inlet condition, speed or discharge pressure. For moisture-bearing air, the applicable dry/wet method and water-vapour treatment must also be established.
Do not apply this conversion to an instrument output already expressed on the required reference basis. Do not use it as a replacement for the agreed compressor test-code correction. If the instrument or supplier documentation is insufficient to determine the basis, preserve the reading but hold the acceptance comparison. Section 20 illustrates this distinction without asserting a formal FAD certification.
9. Preserve pressure, temperature and correction conditions
Record discharge pressure at the location used by the performance requirement. Record the pressure at the flow-meter location separately where a conversion needs it. If gauge pressure is converted to absolute pressure, use the corresponding atmospheric pressure on the same unit basis. A convenient fixed atmospheric value must not impersonate a site measurement when the test needs the actual value.
Temperature conversions use T(K) = t(°C) + 273.15. Room, compressor inlet, aftercooler outlet and flow-meter temperatures can differ. Identify the quantity actually required by the method. An ambient correction derived from one location should not be applied automatically to a reading from another.
For every performance correction, retain the original value, formula or table reference, applicability range, inputs, corrected value and reviewer. A supplier curve correction, an instrument scaling correction and a reference-volume conversion have different purposes. Applying several without recording their relationship can correct the same effect twice.
Where the observed test point lies outside the approved correction envelope, report the measurement as observed and hold the affected acceptance. Do not extrapolate a convenient curve solely to produce the required result. The next step may be a valid retest or an explicitly approved alternative method.
10. Measure electrical input on the agreed package boundary
Use measured true electrical power or interval energy from a suitable measurement arrangement. Motor nameplate power describes a rating; it is not a measurement of package input during the test. Current alone is not sufficient to establish true power. The measurement should include the auxiliaries required by the acceptance boundary and exclude unrelated loads.
List separately supplied fans, pumps, dryer heaters, ventilation and controls. Add a separate load only when it belongs inside the defined boundary and is not already included in the primary feeder measurement. Keep the addition visible so that another reviewer can reconstruct the total.
When a valid energy meter covers the accepted interval, average input power is energy change divided by elapsed hours. For variable-speed equipment, use a method suitable for the waveform and supply-side boundary. Do not treat an unsuitable motor-output reading as interchangeable with purchased package input.
State how cyclic auxiliaries are represented. A short interval during which a cooling fan happens to be off may not represent the duty promised by the comparison. The test plan should determine whether the fan state is controlled, sampled over representative cycling or separately accounted for; it should not be selected retrospectively according to which result is lowest.
11. Use synchronised, valid and representative test intervals
Pair power, flow, pressure and operating-state data on a common clock and a common accepted interval. Retain start and end timestamps, excluded records and the reason for exclusion. Missing readings must not become zero flow or zero power. An alarm, a sensor dropout and an intentional stop are different events with different treatment.
For interval-average readings, time-weighted means are:
- Q_mean = Σ(Q_i × Δt_i) / ΣΔt_i
- P_mean = Σ(P_i × Δt_i) / ΣΔt_i
Use these only for the intervals valid for the comparison. Where readings are instantaneous rather than interval averages, declare the integration or averaging method. An irregular series cannot be averaged by row count and assumed to represent elapsed time.
Do not reduce the dataset to the single best reading. Use the pre-agreed run-combination method, whether it assesses every run separately or combines valid repeats. Preserve rejected runs and distinguish invalid conditions from valid measurements that fail a performance criterion.
A steady mean can conceal an unacceptable excursion. Where a pressure minimum, maximum temperature or control response is itself a requirement, assess that criterion on its defined time basis in addition to the average performance. Averaging cannot erase a failure of an independently specified limit.
12. Calculate performance and margins without hiding the basis
For aligned package input P in kW and reference airflow Q in m³/min, define specific power SP = P / Q, in kW per m³/min. The same numerical relationship can be expressed as specific energy e = P / (60Q), in kWh/m³, when the power and flow represent the same period. The factor 60 is essential because flow is expressed per minute and energy per hour.
For variable operation, use integrated energy divided by integrated reference volume for the period's specific energy. A ratio of aligned time-weighted means gives the corresponding aggregate specific power. It is not generally the same as the arithmetic mean of individual power-to-flow ratios.
Useful descriptive margins are:
- Flow margin (%) = 100 × (Q_result − Q_min) / Q_min
- Power headroom (%) = 100 × (P_max − P_result) / P_max
- Specific-power headroom (%) = 100 × (SP_max − SP_result) / SP_max
These conventions make a positive value favourable, but a margin does not replace the agreed uncertainty decision rule. Display the limit and direction alongside the percentage. If a denominator is zero or absent, suppress the percentage rather than return an error or a misleading infinity.
Never substitute a good ratio for an unmet absolute requirement. A compressor can have acceptable specific power while delivering insufficient flow. Capacity, power, pressure and other mandatory criteria remain separate requirements.
13. Keep tolerance, uncertainty and repeatability distinct
Tolerance or an agreed performance allowance defines what the requirement permits. Measurement uncertainty describes the uncertainty associated with the measured result. Repeatability describes variation under stated repeated conditions. None is a licence to move the acceptance limit after a failed test. The worksheet records them in separate fields and identifies the decision rule that connects the result to the limit.
NIST's uncertainty guidance combines standard uncertainty contributions with sensitivity and covariance terms where appropriate; expanded uncertainty is U = k × u_c, with the coverage factor reported. A coverage factor alone does not justify an unsupported confidence statement. [R2, R3]
This formula assumes the inputs and their uncertainty descriptions are suitable for that approximation. Use standard uncertainties, not an unexplained mixture of certificate limits, percentages and expanded uncertainties. Divide an expanded uncertainty by its stated coverage factor where that conversion is appropriate. The covariance term must use the same units and uncertainty basis.
Repeated observations do not automatically remove calibration, installation or common scaling uncertainty. If a simplified example assumes independence and a complete uncertainty budget, label those assumptions. In a real acceptance, unresolved material contributions require evaluation; they must not be silently set to zero because doing so makes the comparison pass.
14. Agree the decision rule before using pass or fail
The project must state how uncertainty affects conformity. Published guidance recognises decision rules as an explicit part of conformity reporting; this worksheet does not impose one rule on every contract. [R4] A simple limit comparison and an uncertainty-aware comparison may legitimately produce different dispositions, so the selected rule belongs in the requirement record.
The following is an optional three-way worksheet rule, newly defined here for transparent examples. It is not a statement that every compressor test standard requires it. Let y be the accepted result and U its nonnegative expanded uncertainty; use the interval [y − U, y + U]. Assume inclusive limits unless the requirement states otherwise.
Requirement type | Conforming under this example rule | Nonconforming under this example rule | Otherwise |
|---|---|---|---|
Minimum L | y − U ≥ L | y + U < L | Indeterminate |
Maximum H | y + U ≤ H | y − U > H | Indeterminate |
Permitted interval [L,H] | Entire result interval lies within [L,H] | Entire result interval lies below L or above H | Indeterminate |
An indeterminate result overlaps a boundary and requires the agreed disposition; it is not an automatic acceptance or proof of failure. “Hold” is used separately for missing requirements, invalid conditions or unavailable evidence. Where a valid mandatory criterion is nonconforming and another is on hold, retain both findings rather than hiding the known failure.
Conditional release requires an authorised decision specifying scope, limitations and follow-up. It must not overwrite the underlying nonconforming, indeterminate or untested result. More favourable uncertainty treatment must never be selected merely because the first calculation was inconvenient.
15. Test part-load and control behaviour on their own terms
A full-load pass does not demonstrate performance at all operating states. Where specified, record the supported part-load points, control mode, actual speed, flow, package power and pressure stability. Do not infer an entire part-load curve from one full-load reading or extrapolate below the manufacturer's supported operating envelope.
For load/unload equipment, record unloaded power and the relevant state or cycle duration separately. At zero delivered flow, P/Q is undefined; report unloaded kW or cycle energy instead of generating an infinite specific-power value. For variable-speed equipment, identify transitions to unloading, stopping or another control mode rather than treating every low-demand state as stable speed control.
Functional acceptance should describe the actual response to an approved demand or control event. Include timing, pressure excursion, recovery, interaction with other machines and the agreed response criterion where those are in scope. These are not invitations to create unsafe trips or disable safeguards; the test method and operating permissions must be established separately.
A complete-cycle average may be useful for energy performance, but it cannot replace a peak response or minimum-pressure criterion that applies within the cycle.
16. Reconcile net delivery, purge and receiver inventory
A flow meter downstream of treatment may measure useful air delivery, while the compressor guarantee applies upstream. Purge, other outflows and receiver charging or discharging can account for a difference. That difference should be reconciled on one gas-quantity and reference basis before it is attributed to the compressor.
A positive storage change means air accumulates inside the boundary, reducing the volume crossing the net outlet. A negative storage change means inventory is being discharged and can temporarily make downstream delivery exceed current compressor production. Use a defensible inventory calculation with the required absolute pressures, temperatures, effective volume and gas basis; a pressure difference alone is not a volume measurement.
Do not add a purge allowance if the selected meter already measures gross flow or subtract it twice from a net reading. Do not count another compressor's contribution as output from the unit under test. Where interactions cannot be measured or controlled well enough, move the measurement boundary or arrange a suitable retest.
The same distinction applies to power. Compressor-package specific power and treated-system specific power can both be useful, but they describe different boundaries and should be compared with their own requirements.
17. Keep air quality, mechanical condition and readiness visible
A capacity-and-power result does not establish every required property of the compressed air. If dew point, particles, oil or another quality parameter is specified, identify the parameter, sampling location, pressure, method and applicable conditions. A technology label or visually clear condensate cannot substitute for the required evidence.
Post-overhaul condition checks may include leakage, temperature, vibration and other defined observations linked to the actual repair scope. Record relevant load, speed, ambient and cooling conditions. A low temperature immediately after startup does not prove thermal stability; a quiet observation does not prove a universal vibration limit.
Track required control, protection and standby functions through their own approved tests. Keep untested functions visible rather than assigning them a pass from another result. A package may meet its airflow requirement while the installed system still has an unresolved treatment or critical-user-pressure issue.
Where full site duty cannot yet be demonstrated, describe the achieved scope and missing condition precisely. Do not replace missing peak-load evidence with a statement that all operating conditions passed. Release for limited operation and full performance acceptance are different decisions and should remain distinguishable in the handover.
18. Worked example A: aligned full-load acceptance
All figures in Sections 18–21 are synthetic teaching and calculation cases, not supplier guarantees or customer tests. In this case the fictional requirement sets minimum reference airflow of 20.00 m³/min, maximum package input of 126.0 kW and maximum specific power of 6.20 kW per m³/min. The agreed pressure condition is satisfied, the reference basis matches and three equal-duration stable runs may be combined under the fictional plan.
Run | Reference airflow, m³/min | Package input, kW | Treatment |
|---|---|---|---|
A1 | 20.40 | 123.0 | Valid stable interval |
A2 | 20.30 | 124.0 | Valid stable interval |
A3 | 20.50 | 122.0 | Valid stable interval |
Combined mean | 20.40 | 123.0 | Equal-duration average |
The ratio of the means is 123 / 20.4 = 6.0294 kW per m³/min. Specific energy is 6.0294 / 60 = 0.10049 kWh/m³. Flow margin is +2.00%; power headroom is +2.38%. These are descriptive results before the decision rule is applied.
Assume the complete uncertainty assessment for these means gives standard uncertainties u_Q = 0.125 m³/min and u_P = 0.50 kW, with independence and coverage factor k = 2. Thus U_Q = 0.25 m³/min and U_P = 1.00 kW. First-order ratio propagation gives u_SP = 0.04434 and U_SP = 0.08867 kW per m³/min.
Under Section 14's example rule, the lower airflow bound is 20.15, above 20.00; the upper power bound is 124.0, below 126.0; and the upper specific-power bound is 6.1181, below 6.20. These three numerical criteria conform. Final acceptance still depends on every other mandatory requirement and required handover item. The example does not claim that independent input uncertainty or k = 2 is appropriate for every real installation.
19. Worked example B: a positive margin can remain indeterminate
A second fictional test has the same 20.00 m³/min minimum airflow requirement. A valid measured result is 20.10 m³/min with expanded uncertainty U = 0.25 m³/min. The raw margin is positive, but the result interval spans 19.85 to 20.35 m³/min. Under the stated three-way rule, this overlaps the minimum and is indeterminate.
Result | Expanded uncertainty | Result interval, m³/min | Disposition against minimum 20.00 |
|---|---|---|---|
20.10 | 0.25 | 19.85–20.35 | Indeterminate |
19.60 | 0.25 | 19.35–19.85 | Nonconforming |
20.40 | 0.25 | 20.15–20.65 | Conforming |
20.40 | Not established | Not established | Hold uncertainty-aware decision |
The second row is entirely below the minimum; the third is entirely above it. The fourth must not silently assume zero uncertainty. A different contractually agreed decision rule could produce another result, but changing rules after seeing the data is not part of this example.
Now suppose a nominal requirement of 20.00 m³/min explicitly permits a 3% negative capacity allowance. Its resulting minimum would be 19.40 m³/min, if that is genuinely how the governing requirement is written. The allowance must be incorporated once, as a separate requirement version, not combined with an extra unapproved 3% after comparison. This hypothetical allowance is not a statement of any industry-standard tolerance.
The proper next action for the indeterminate result may be a better-supported measurement, an agreed retest or a documented contractual disposition. Repeating readings while ignoring a systematic installation error will not necessarily resolve it.
20. Worked example C: reference conversion is not a capacity correction
A hypothetical meter reports actual dry-air flow of 2.500 m³/min at 8.00 bar absolute and 40.00°C. The requested volume basis is dry air at 1.00 bar absolute and 20.00°C. Assume unchanged composition, equal compressibility factors, no unaccounted flow and measurements valid for this elementary conversion.
Q_ref = 2.500 × (8.00 / 1.00) × (293.15 / 313.15)
Q_ref = 18.7227 m³/min on the stated reference basis
Using 7.00 bar gauge incorrectly in place of 8.00 bar absolute would give 16.3823 m³/min—12.5% below the correctly converted value. The calculation shows why the pressure basis is a required field, not a label added after processing.
Calculation state | Result | Correct use |
|---|---|---|
Actual dry flow at the stated meter conditions | 2.500 m³/min | Preserved source quantity |
Converted to the stated reference volume | 18.7227 m³/min | Comparable only with the same reference basis |
Gauge pressure incorrectly substituted | 16.3823 m³/min | Invalid conversion; retain as a diagnosed error |
If the real instrument already outputs 18.7227 reference m³/min, applying the pressure-temperature conversion again would be double correction. Confirm the configured output type before calculating.
This case does not establish what airflow the compressor would deliver at another discharge pressure or another inlet condition. Nor does it prove formal test-code compliance. Those questions require the relevant performance method and evidence. The worksheet may show a correct volumetric conversion while keeping the compressor acceptance on hold because the actual guarantee uses a different test basis.
21. Worked example D: package output is not treated-system output
Use Example A's 20.40 reference m³/min and 123.0 kW as the compressor-package observations. During a fictional 15-minute interval, the wider treatment-and-storage boundary has average purge of 1.20 reference m³/min, other recorded outflow of 0.10 reference m³/min and positive inventory accumulation equivalent to 0.20 reference m³/min. There is no other air source, and all terms share one reference basis.
Net delivery is 20.40 − 1.20 − 0.10 − 0.20 = 18.90 reference m³/min. The interval delivers 283.5 reference m³. Separately supplied treatment power is 4.0 kW and an included cooling auxiliary consumes 2.0 kW; neither is already in the package measurement. Wider-system input is therefore 129.0 kW, or 32.25 kWh over 15 minutes.
Boundary metric | Package | Wider treated system |
|---|---|---|
Airflow | 20.40 m³/min | 18.90 m³/min |
Included electrical input | 123.0 kW | 129.0 kW |
Specific power | 6.0294 kW per m³/min | 6.8254 kW per m³/min |
Specific energy | 0.10049 kWh/m³ | 0.11376 kWh/m³ |
The wider ratio is approximately 13.20% higher, but that does not prove package deterioration: the boundary changed. Both results must carry their own definitions.
Suppose the system separately requires at least 19.20 reference m³/min at its delivery boundary and its complete expanded flow uncertainty is 0.15 m³/min. The upper bound is 19.05, below 19.20, so that system criterion is nonconforming under the example rule even though the package criteria in Example A conform. Investigate the stated system boundary and obligations rather than automatically blaming the airend or treating the package pass as a system pass. Retain the separate inventory term; do not silently add it back unless the approved test method permits that treatment for the criterion being assessed.
22. Editable worksheet A: acceptance definition and responsibility
Complete this record before collecting acceptance data. Replace the blank entries with project information and duplicate requirement rows as necessary. These Word cells are manually editable; they do not execute calculations. A record can be useful during planning while remaining clearly marked as unapproved.
A complete-looking table is not an approval. Identify whether the record is draft, agreed for test or superseded. Preserve the version that was in force during the run. If a requirement changes later, keep the original result and issue a traceable reassessment under the revised requirement instead of editing the historical limit in place.
23. Editable worksheet B: witnessed run and calculation record
Use one record per test point and valid run, with linked raw data. Keep original observations distinguishable from corrections and derived values. A compact summary should allow the reviewer to locate the full interval, not replace it.
Do not use a single unrestricted note field as the only location for a material correction. Record which values were transformed and why. An interrupted run must retain its event record even when the next run succeeds. Where one measurement is unavailable, state which derived result cannot be supported rather than populating the blank with a typical value.
24. Editable worksheet C: requirement disposition and release
Assess each mandatory requirement independently and then summarise the complete scope. Separate the algorithm's numerical disposition from authorised acceptance. This keeps a technical nonconformance visible when a commercial or operational decision permits a limited next step.
The summary should name every unresolved mandatory item. A result such as “three numerical criteria conform; quality evidence pending” is more informative than “75% accepted.” Do not average acceptance percentages across unrelated requirements. Missing quality evidence is not compensated by extra airflow, and an excellent specific-power ratio does not cancel an unsafe installation condition.
25. Correct, retest and preserve the accepted state
When a valid requirement is not met, preserve the failed run, conditions and calculations before changing equipment or settings. Diagnose the relevant chain: instrument configuration, intake, treatment restriction, cooling, valves, controls, drive and other system influences may require different investigations. The result identifies an unmet requirement; it does not automatically identify its cause.
A retest record should connect the original finding to the actual corrective action and the new test. Repeat the affected scope under agreed conditions and assess related requirements that the intervention could change. A speed increase might improve airflow while affecting power, thermal behaviour or the supported operating envelope. Rechecking only the originally failed number can miss a new problem.
Maintain separate dates for completion, testing, acceptance and operational release. Preserve final settings, installed configuration, accepted deviations, raw data, calculation revisions and remaining punch items. A later comparison should be able to distinguish original underperformance from changed duty, maintenance condition or configuration drift.
For an acceptance-review enquiry, provide the agreed performance sheet, actual equipment identity, test plan, instrument details, raw readings and unresolved requirements. The technical scope and responsible reviewer can then be established without inventing guarantees or claiming that this blank worksheet is itself an acceptance certificate.
26. Questions from owners, engineers and commissioning teams
Does reaching the required pressure prove compressor capacity?
No. Pressure is one condition of the test. Capacity needs a suitable flow measurement or accepted test method on the required reference basis. Stored air or another connected compressor can maintain pressure while the unit being assessed provides less flow than expected.
Can I use motor nameplate kW for the power acceptance?
Not as a substitute for measured input. The requirement must identify whether it concerns motor, package or wider-system input. Measure that boundary and account for included auxiliaries without adding loads already captured by the main meter.
Are factory acceptance and site acceptance interchangeable?
No. Their scopes differ. Factory evidence can demonstrate agreed package requirements; installed-system evidence must address actual project interfaces and site conditions. Retain both where required and identify which requirement each closes.
Does a CAGI sheet prove that our installed compressor has passed?
It can support the model's stated performance comparison, with programme status and scope checked. It does not replace the project's required test of the delivered configuration, installed interfaces or site-specific obligations. [R1]
Can a tolerance and measurement uncertainty both be considered?
Yes, when their different roles and the decision rule are explicit. The requirement establishes the permitted limits; uncertainty is treated through the agreed decision rule. Do not enlarge a limit twice or select the treatment after seeing which method produces a pass.
What happens when the uncertainty interval crosses a limit?
Under the optional rule in this worksheet, the result is indeterminate. The project then follows its agreed next step. Another documented rule may apply in the contract, but it must be identified before its result is used. An indeterminate result is not a permanent equipment diagnosis.
Should a zero-flow run have a specific-power result?
No. Division by zero is undefined. Record unloaded or standby power and relevant cycle energy separately. A zero-flow observation may be valid for that test purpose, but it cannot support a numerical power-per-delivered-flow ratio.
Can a downstream meter assess a package airflow guarantee?
Only when its relationship to the package boundary is properly established through an approved method. Purge, other inflows/outflows and storage change can make downstream delivery differ from package output. Do not silently equate the two.
Can we change the acceptance limit after an unsuccessful test?
Not silently. Preserve the result and any proposed change, with the responsible technical and contractual approval. Reassessment under a legitimately changed requirement must remain traceable to both versions.