Before and After Comparison Method

Before and After Comparison Method

OWNED DECISION: Establish whether a before-and-after compressed-air comparison is comparable, uncertainty-aware and fit for a specific public claim. MANDATORY ASSET: Project Comparison Methodology.

1. Direct engineering answer

A defensible before-and-after comparison defines the same physical and operational boundary, documents what changed, records baseline and post-intervention duty, aligns measurement periods, normalises material differences, quantifies uncertainty and excludes rival causes. The result is a qualified difference at a defined boundary—not automatic proof of savings, capacity gain, reliability improvement or customer production benefit.

2. Distinct ownership

3. First define the claim

Before collecting data, write the proposed claim as a testable statement: “station electrical energy per matched production period declined”, “pressure at the critical point improved at comparable demand”, or “dew point at the named sampling point met the agreed threshold during the observation window”. Each needs a different measurement plan.

4. Boundary map

Draw the air path and electrical boundary. Mark compressor package, auxiliary loads, dryers, filters, receivers, header, distribution, critical users and any parallel sources. Mark whether the claim concerns a machine, station, branch or process. A package-only reading cannot establish whole-station performance.

5. Baseline selection

Choose a baseline period representing normal operation and sufficient duty variability, rather than the worst historical hour. Record dates, shift mix, output, seasonal conditions, maintenance state, leaks, control settings and exceptional events. Retain the raw baseline even when adjustments are subsequently approved.

6. Post-intervention period

Define the intervention completion date, commissioning/stabilisation interval and measurement start. Exclude start-up transients only under a predeclared rule. Do not compare a peak-demand baseline with a low-demand post-intervention shift and call the resulting energy difference an efficiency gain.

7. Physical change register

Record every intervention: piping alterations, leak repairs, compressor replacement, setpoint changes, dryer servicing, storage changes, production modifications and controls updates. If multiple changes occurred, the evidence may support a combined project effect but not the isolated benefit of one component.

8. Match duty before comparing

At minimum examine delivered-air flow, demand profile, pressure setpoint, equipment states, operating hours, production output, ambient conditions, treatment load and quality requirements. A single matched point is useful for component diagnosis but may not represent annual station energy.

9. Comparison hierarchy

Preferred: simultaneous or tightly matched duty with stable measurement boundaries. Next: repeated representative operating states with documented weighting. Next: regression or engineering-model normalisation validated across the observed range. Lowest confidence: unadjusted bill or single-point comparison with material confounders. Choose method based on claim and evidence.

10. Station versus machine efficiency

Machine-specific power can improve while station kWh rises if utilisation, dryer load or auxiliary operation changes. Conversely station consumption may fall because production dropped. State which level is being compared, the corresponding flow reference and the demand distribution.

11. Electrical energy baseline

Electrical energy E = integral of true input power P(t) over time; for interval data E ≈ sum(Pi × Δti). Document meter placement, auxiliaries, CT ratios, sampling interval, gaps and treatment of standby. Motor nameplate power or current alone is not a verified energy baseline.

12. Flow and air-delivery basis

Document measured volumetric flow and reference conditions; distinguish actual line volume from referenced delivery. For a flow-normalised comparison, use the same volume reference, pressure, temperature basis and instrument correction method. State whether air delivered to useful end uses or gross generated air is measured.

13. Pressure basis

Use comparable pressure tap locations, gauge/absolute conventions and synchronous demand states. A reduction in compressor discharge pressure is not sufficient if critical end-use pressure deteriorates. Record the pressure band and transients, not just a daily average.

14. Quality basis

If the intervention changes dryer or filter operation, compare pressure dew point, particle or oil measurements at equivalent sampling locations and duty. A lower energy result is not an acceptable improvement if required compressed-air quality is compromised.

15. Normalisation principle

Normalisation adjusts an observed outcome to a defined reference operating condition, not to an arbitrary favourable scenario. Specify variables, model form, training period, valid operating range, residual behaviour and who approved the adjustment. Retain both observed and adjusted results.

16. Simple production normalisation

Illustration: kWh per accepted production unit can be informative if output is a credible driver of air demand. It can mislead when substantial base load, idle air, mixed product types or non-air electrical loads remain. Record the denominator, product mix and nonlinearity.

17. Demand-weighted specific power

For a duty profile, assess station kW and delivered flow across operating bins, with each bin weighted by observed time or air delivery as appropriate. Avoid averaging individual specific-power ratios without explaining the weighting. The ratio of total kWh to total referenced delivered-air volume is often more defensible for a measured period.

18. Regression normalisation

A model such as energy = a + b×production + c×operating hours may be useful only if predictors are technically justified, sufficiently varied and measured consistently. Inspect residuals, outliers, collinearity and extrapolation.

19. Seasonal and ambient adjustment

Ambient temperature, humidity and cooling conditions can affect compressor and dryer operation. Adjust only when evidence shows material influence. Use logged ambient values and validated relationships rather than assuming a universal correction coefficient.

20. Leakage and artificial demand

Leakage repair can reduce air demand and energy, but pressure setpoint and control behaviour influence the realised electrical effect. Document compressor loaded/unloaded/VSD response, simultaneous plant demand and whether reduced leakage changed useful production supply.

21. Controls and sequencing

A sequencing change can alter idle hours and compressor mix without changing process demand. Compare total station input at matched demand and pressure; log individual machine states. Do not extrapolate from one stable shift to a year without the annual duty profile.

22. Treatment and distribution interventions

Filter or piping changes can lower differential pressure at a given flow. Whether station kWh falls depends on downstream pressure requirement and upstream control response. State the measured hydraulic benefit separately from any verified electrical benefit.

23. Confounder register

Common confounders: production volume/mix, shutdowns, equipment additions, maintenance, ambient conditions, changed pressure target, air leaks, parallel compressors, sampling position, meter replacement, quality requirements and operator behaviour. Each requires assessment, not merely a disclaimer.

24. Instrument continuity

When instruments change between periods, establish comparable ranges, reference conditions and corrections. An unverified meter swap may invalidate a small claimed effect.

25. Clock and data alignment

Align electrical, flow, pressure, production and control-state records on a documented timebase. Preserve native sample resolution and data gaps. Do not interpolate short events into a false continuous record. State whether averages, medians or percentile bands are used and why.

26. Missing data policy

Set completeness and exclusion rules before calculating outcomes. Identify communication gaps, sensor saturation, invalid operating states and maintenance intervals. Report the proportion of excluded data and test whether omissions disproportionately remove poor performance.

27. Uncertainty of a difference

For independent standard uncertainties u_before and u_after, u_difference = sqrt(u_before² + u_after²). Correlation may increase or decrease this uncertainty; shared systematic errors require covariance treatment. Field variability and model error must be considered in addition to certificate figures.

28. Confidence and practical significance

A positive observed difference is not necessarily distinguishable from noise. Report uncertainty, sample size, range of operating states and practical magnitude. A statistically distinguishable change may still be operationally unimportant, and vice versa.

29. Energy-to-cost conversion

Verified energy difference may be converted to cost only with applicable tariff structure, time-of-use periods, demand charges and other relevant terms. Avoid multiplying an isolated kW reduction by 8,760 hours unless equivalent operation across those hours is supported.

30. Carbon comparison

For emissions reporting, document location-based or market-based method, electricity factors, reporting period, source date and accounting boundary. Do not call a modelled kWh saving an independently verified carbon reduction.

31. Attribution ladder

Level A: observed numerical difference. Level B: difference after documented comparability checks. Level C: uncertainty-qualified improvement. Level D: improvement plausibly associated with the intervention after evaluating confounders. Level E: independently supported causal attribution under a fit-for-purpose design. Never promote a Level A record to Level E by editorial wording.

32. Acceptance categories

ACCEPT: consistent boundaries, periods, appropriate method, uncertainty and reviewed claims. QUALIFY: narrow result supportable with explicit limitations. REWORK: normalisation or meter evidence incomplete. REJECT: incompatible boundaries, invented data, material unassessed confounders or absent publication consent.

33. Project timeline and approval

The public case study should not precede acceptance of the underlying evidence or customer permission.

34. Evidence retention

Preserve native logger files, meter certificates, operating logs, intervention work orders, calculation workbook versions, analysis code if used, model parameters, approval decisions and redacted public extracts. Corrections must be versioned rather than silently overwriting results.

35. Publication integrity

Label every worked numerical example as illustrative unless a customer-backed evidence pack is supplied. Do not infer project location, OEM endorsement, performance warranty, guaranteed savings or a verified customer outcome from educational calculations.

36. Safety and technical scope

Measurements at pressurised equipment, live electrical systems or contamination-sensitive points require competent personnel and site-specific safe systems of work. This editorial methodology is not a commissioning instruction or a substitute for engineering sign-off.

37. Applicable method families

Potential references include ISO 11011 for compressed-air energy assessment, ISO 50015 and IPMVP for measurement and verification, ISO 8573 for specified air-quality measurement, and ISO/IEC 17025 for calibration laboratory competence. Technical owners must verify edition, scope, suitability and any contract-specific requirements; these references are not certification claims.

38. Editorial boundary with

39. Authoritative source integration

Supplied B.A. (11-8-26) RC.pdf discusses specific power and the need to assess performance beyond equipment availability; B.A. (16-8-26) RC-2.pdf stresses that specific power must be interpreted in context; B.A. (04-8-26) RC-5.pdf includes measurement-oriented material.

40. Final technical position

41. Worked comparison A — energy and production mix

ILLUSTRATIVE ONLY. Baseline: 1,000 kWh over a shift producing 100 units; post: 880 kWh producing 80 units. Unadjusted energy fell 120 kWh (12%), but energy per unit rose from 10 to 11 kWh/unit (10%). Neither ratio alone establishes compressor efficiency because production mix, fixed station load, air delivery and other electric loads are unknown. Disposition: no efficiency claim; obtain station-only kWh, delivered air and matched production context.

42. Worked comparison B — matched flow and specific power

ILLUSTRATIVE ONLY. At the same referenced delivery of 12 m³/min and pressure target, baseline station input is 96 kW and post input is 88 kW. Specific power changes from 8.00 to 7.33 kW/(m³/min), a measured point difference of 0.67 kW/(m³/min) or 8.33% relative to baseline. Confirm station auxiliary boundary, demand stability, quality and meter uncertainty. This supports only a matched-point observation, not annual kWh savings.

43. Worked comparison C — pressure and confounders

ILLUSTRATIVE ONLY. A critical branch reads 5.8 bar(g) before and 6.2 bar(g) after a pipe change. But pre-test flow is 14 m³/min and post-test flow is 10 m³/min. The pressure difference cannot be assigned to the pipe change without flow matching or a validated hydraulic model. Repeat at comparable demand and record upstream pressure and control state.

44. Worked comparison D — annualisation risk

ILLUSTRATIVE ONLY. A verified 8 kW reduction at a specific production state is observed for 2 hours. The measured energy difference is 16 kWh for that observation. Multiplying 8 kW by 6,000 annual operating hours would yield 48,000 kWh only if that operating state and reduction persisted across those hours; absent representative duty evidence, 48,000 kWh is a scenario, not a verified saving.

45. Worked comparison E — uncertainty threshold

ILLUSTRATIVE ONLY. Baseline differential pressure = 0.34 bar; post = 0.28 bar; observed reduction = 0.06 bar. If each reading has standard uncertainty 0.03 bar and independence is justified, standard uncertainty of difference is 0.042 bar; illustrative expanded uncertainty with k=2 is about 0.085 bar. The observed 0.06-bar change is not robust against that stated uncertainty. Disposition: qualify or remeasure with more suitable instrumentation.

46. Worked comparison F — weighted duty profile

ILLUSTRATIVE ONLY. Baseline station power at low/high demand is 40/100 kW; post is 35/90 kW. At 70% low and 30% high duty, baseline mean power = 58 kW and post = 51.5 kW, difference 6.5 kW. At a different 30% low and 70% high duty, baseline = 82 kW and post = 73.5 kW, difference 8.5 kW. Weighting must follow a declared representative duty profile; choosing the more favourable mix is not acceptable.

47. Worked comparison G — model boundary

ILLUSTRATIVE ONLY. A baseline regression predicts 120 kWh for the observed post-period production and hours; measured post energy is 108 kWh. Model-adjusted difference is 12 kWh, but the claim requires prediction uncertainty, residual validation, evidence that the post-period conditions fall within the model training range and checks for concurrent interventions. Without these, publish only the illustrative method.

48. Worked comparison H — contamination control

ILLUSTRATIVE ONLY. Pressure dew point at a sampled header improves from +8 °C to −18 °C under stated pressure conditions. A downstream critical user was not sampled. The record may support an improvement at the header during the tested windows, but not a plant-wide air-quality certification or a permanent point-of-use guarantee.

49. Worked comparison I — leak repair and control response

ILLUSTRATIVE ONLY. Ultrasonic survey and follow-up indicate repaired leak sources, while station electrical energy over two non-matched shifts falls 9%. Because production demand and compressor control modes changed, 9% cannot be credited to leakage repair. Match demand, log compressor loaded/unloaded states and calculate the supported station-level effect.

50. Worked comparison J — parallel supply

ILLUSTRATIVE ONLY. Main compressor kWh declines after intervention, but a temporary rental unit supplies part of the air demand. Comparing only the main compressor understates the total supply boundary. Include the temporary unit and its fuel/electric energy on a consistent accounting basis or narrow the claim to main-unit input.

51. Mandatory asset — Project Comparison Methodology

Identity and governance

Scope and timeline

Operating comparability

Measurement evidence

Analysis and disposition

52. Comparison decision workflow