Measured Sequencing Optimisation Project Records

Measured Sequencing Optimisation Project Records

1. Direct engineering answer

A measured sequencing optimisation project record shows how a compressed-air station dispatched individual compressors before and after a defined control intervention, under comparable demand, pressure and production conditions. The proof is a synchronised record of machine states, delivered-air demand, station electrical input, pressure performance, intervention dates, exceptions and uncertainty—not merely a claim that a master controller was installed. A credible publication separates observed dispatch changes from measured energy changes, modelled counterfactuals and unsupported causal claims.

The owned decision is whether the project record documents representative demand and changes in machine dispatch performance well enough for an evidence-led claim.

2. Controlled scope and adjacent-page boundaries

Distinguish control-system design guidance from an evidence-dependent project narrative. A control logic diagram is not a verified field result. A single load/unload screenshot cannot establish shift-wide behaviour. A vendor simulation is not measured savings. The project record must state which of these it actually contains.

3. System boundary and machine inventory

Draw the compressed-air boundary from each compressor electrical incomer and package outlet to common receivers, treatment, headers and the nominated demand meter. Identify parallel plants, backup units, rental equipment, interconnecting valves and any export/import flow. Record whether measured power includes cooling fans, dryers, drains, compressor-room ventilation or only compressor package electrical input. Mixing these boundaries before and after invalidates a direct comparison.

4. What sequencing changes in engineering terms

Base and trim assignment

Base compressors carry sustained demand while a trim compressor responds to variation; however, base/trim labels alone do not prove efficiency. Verify each unit’s part-load performance and actual station operating profile. A variable-speed unit may be efficient within a certain control envelope but inefficient or constrained elsewhere.

Load/unload interaction

Two machines can both unload or hunt when pressure bands and communications are poorly coordinated. Record loaded, unloaded, stopped and fault time separately, with controller definitions and actual measured power for each state.

Start/stop and minimum-run constraints

Sequencing must respect OEM permitted starts per hour, minimum run times, unloaded cooldown, thermal behaviour and maintenance limitations. No universal values are supplied here; the applicable serial-specific documentation and qualified owner decide.

VSD control range and trim saturation

A VSD trim may hit minimum or maximum speed, causing a second unit to load or rapid cycling. Preserve VSD speed, motor power, output pressure and demand traces, and note controller override events.

Pressure band and storage

Receiver volume and allowed pressure band determine short-term buffer capacity, while actual temperature, compressibility and operating pressure affect usable air. Sequence evaluation must preserve critical end-use service.

Machine rotation and maintenance

Lead/lag rotation may distribute running hours, but it can also shift dispatch toward a less efficient unit. Document why the rotation schedule was chosen, and keep maintenance and availability benefits separate from kWh claims.

Communication loss and fallback

Capture controller failure, sensor failure, network interruption, manual override and local fallback behaviour; commissioning approval belongs to the responsible controls/OEM and plant owners.

Quality and process protection

A dispatch change can alter air temperature, moisture carryover, pressure ripple and dryer operating conditions. Record treatment status and quality acceptance where process-sensitive; do not imply that kW reduction alone demonstrates suitability.

5. Measurement architecture and synchronisation

For each compressor record true power (kW), energy (kWh), operating state, load percentage or VSD speed, outlet pressure, alarms, running hours and controller command. For the station record delivered-air flow, common-header pressure, receiver pressure, relevant treatment differential pressure, critical inlet pressure and production/demand context. Use a single documented time reference; reconcile meter clock drift and controller event latency. Record raw sample interval and aggregation method.

Electrical input should be measured by suitable calibrated instruments and proper electrical safety procedures. A controller’s calculated power may be useful for trend analysis but should not silently replace a verified true-power meter. Compressor state tags can mean commanded rather than actual loaded operation; reconcile them with power and flow. Flow meters require stated location, units, standard/reference conditions, pressure, temperature and uncertainty. Do not sum incompatible nominal compressor capacities as a substitute for measured delivered demand.

6. Define representative demand before changing controls

A baseline should cover normal shifts, low/high production, break periods, batch changes, simultaneous peak users, cleaning cycles and any planned shutdowns. Record output quantity, product mix, line utilisation, shift hours and exceptional compressed-air events. State which time windows are excluded and why. An intervention test at low weekend demand cannot prove an annual weekday saving without an accepted demand model and uncertainty.

Pre-register comparable before/after windows and matching criteria where feasible. Include total demand distribution, peak frequency, compressor availability, air quality, leak-repair status, treatment maintenance, pressure target and ambient conditions. A control retrofit performed alongside major leak repair, filter replacement and production restructuring cannot have all observed energy reduction attributed to sequencing.

7. Measurement definitions and KPIs

When flow is in m³/min, kW ÷ (m³/min) has units kW per (m³/min). Never call it kWh/m³ without converting time and verifying the same boundary. Report pressure in bar(g) or kPa(g), not ambiguous “bar”. State whether kWh comes from cumulative meter registers or numerical integration of sampled power.

8. Before/after protocol and intervention trace

8.1 Freeze the baseline

Archive control setpoints, programme version, machine availability, meter setup and a dated operating schematic.

8.2 Record control intent

Document target lead/lag order, pressure bands, VSD trim priority, start/stop criteria and fallback logic, with responsible engineering signoff.

8.3 Execute approved change

Record who changed each parameter, exact timestamp, firmware/configuration backup and any machine taken offline.

8.4 Verify functions

Observe low demand, peak demand, transitions, fault handling, pressure recovery, local overrides and communications failure under an approved test plan.

8.5 Capture comparable after period

Keep same instrument boundary and representative production windows; document deviations.

8.6 Analyse uncertainty and attribution

Reconcile missing data, sensor drift, flow boundary, process mix and concurrent interventions.

8.7 Recheck persistence

Repeat a representative sample after operators and maintenance teams have resumed normal control.

8.8 Approve bounded claims

Separate measured facts, adjusted estimates and modelled savings; record consent, relationship disclosures and editorial signoff.

9. Mandatory asset — Sequencing Project Measurement Narrative

The following is the required controlled project-evidence asset. Each field must have a source or explicit “not available”, an accountable owner and a disposition. Do not use blank fields as evidence of acceptance.

Workflow status for every row: NOT RECEIVED / RECEIVED / VERIFIED / EXCEPTION / REJECTED / NOT APPLICABLE, with method, result, evidence ID, accountable owner, open item and disposition. A project card may publish only approved claims linked to this record. The blank template is a deliverable; no actual customer project has been entered.

10. Evidence chain and claim classification

For publication, every number must carry the site or anonymised scope, measurement interval, units, source reference, uncertainty and status. The record owner must preserve both positive and adverse outcomes. Do not selectively remove pressure excursions, high start counts or operator overrides to create a favourable story.

11. Worked case A — Unloaded running energy (illustrative)

Suppose one compressor’s verified average unloaded electrical power is 29 kW. During an otherwise comparable four-hour illustrative baseline it unloads for 92 minutes, while after control changes it unloads for 26 minutes. Unloaded-energy contributions are 29 × 92/60 = 44.47 kWh and 29 × 26/60 = 12.57 kWh, a difference of 31.90 kWh for that state alone. This is not the station energy saving: the other compressors may run more, and pressure, demand and starts may change. Confirm total metered station kWh and all operating states before making an outcome claim.

12. Worked case B — Total station energy versus production (illustrative)

A fictional baseline interval uses 1,260 kWh for 840 finished units; the after interval uses 1,170 kWh for 900 units. Absolute interval energy differs by 90 kWh, but production differs. Simple ratios are 1.500 and 1.300 kWh/unit respectively. This does not isolate compressor sequencing because the figures could include different machine utilisation or unrelated process energy; only compressor-boundary meters, air demand and appropriate adjustment can support a sequencing-specific interpretation.

13. Worked case C — Control-state overlap (illustrative)

Two fixed-speed machines show 36 minutes of simultaneous unloaded operation in a 180-minute trace. After revised pressure-band coordination, the overlap falls to 8 minutes. The measured state overlap decreases by 28 minutes. If unloaded input powers are 21 and 25 kW, it is invalid to multiply 28 minutes by 46 kW and call the product saved energy unless the counterfactual machine states and the entire station’s actual power are established. A second compressor might be loaded instead of unloaded during part of the period.

14. Worked case D — Trim saturation (illustrative)

A VSD compressor reaches its documented maximum controllable output for 17 minutes during a peak production event; the revised sequence starts a fixed-speed support machine earlier and maintains critical inlet pressure. Total electrical consumption for that event could rise even though pressure reliability improves. Report the dispatch benefit and energy effect separately. The appropriate claim may be “reduced pressure excursions during tested peak demand”, not “lower energy use”.

15. Worked case E — Meter uncertainty near a claimed saving (illustrative)

Two independent period-energy measurements each have illustrative standard uncertainty of 8 kWh. Their difference has combined standard uncertainty sqrt(8²+8²) = 11.31 kWh, before production-adjustment uncertainty. A measured difference of 9 kWh would be smaller than this illustrative combined standard uncertainty. It should not be advertised as a demonstrated energy improvement. Real instruments may have correlated error, and a proper uncertainty budget must reflect it.

16. Worked case F — Missing data and a false improvement (illustrative)

The baseline power meter records every minute, but the after period misses 38 minutes during a high-demand shift. Filling missing after values with the period average could bias energy downward. Retain a gap map, compare controller events, reconstruct only with a defensible method and report the resulting uncertainty; otherwise exclude the period. Never erase the missingness in public charts.

17. Worked case G — Annualisation boundary (illustrative)

A representative controlled trial estimates 12 kWh lower station energy over one documented eight-hour production window. Multiplying by 3 shifts and 300 days would yield 10,800 kWh/year mathematically, but only if the same demand, equipment availability, operating hours and control performance recur. Present this as a conditional scenario, not realised annual savings, and show sensitivity to actual hours and demand profiles.

18. Worked case H — Persistence and drift (illustrative)

At commissioning the new sequence reduces simultaneous unload events. Six weeks later an operator places the trim compressor in local manual mode and the original overlap returns. The initial functional test remains a valid dated observation, but a “persistent savings” claim fails. The evidence narrative must include control overrides, the date of drift, corrective action, subsequent verified check and any revised publication statement.

19. Causality and confounder ledger

20. Normalisation and counterfactual design

Use a matched-window approach when production and demand conditions are genuinely similar. For variable sites, define an adjustment model using measured flow/demand, operating hours, product family and other relevant explanatory variables; retain training/validation windows, residual diagnostics and extrapolation limits. Report whether baseline model predictions are compared with observed after-period kWh. Avoid using post-intervention variables that themselves respond to the intervention without explaining the causal interpretation.

A defensible model may calculate adjusted savings as predicted baseline energy at after-period independent conditions minus measured after energy, with uncertainty. The model must not be labelled an internationally certified method simply because it resembles an energy measurement-and-verification framework. Apply any named method only after its scope, edition and qualified reviewer are verified.

21. Pressure, reliability and process acceptance

Sequence optimisation can lower kWh yet fail operational acceptance if a critical tool inlet falls below its approved envelope during transients. Record critical inlet minima, duration, alarms, production quality, compressor trips, starts/hour and thermal limits. Include explicit rollback criteria approved by the plant and OEM/controls owner. Do not prescribe a universal pressure band, cycling limit or number of permitted starts.

A successful evidence narrative may report improved pressure stability with neutral or higher energy consumption, or reduced energy with unchanged process quality, provided each claim is independently measured and approved. Report the actual trade-off instead of forcing all outcomes into a savings headline.

22. Instrument, data and security governance

Retain raw meter exports, PLC/controller event logs, control backups, meter configuration, instrument certificates, consent records, signed acceptance and chart-generation code or method. Apply read-only source preservation and a revision log. Redact customer identity, sensitive plant layouts, controller network addresses and proprietary configurations before public release. If customer permission is withdrawn, suspend the public case and document cache, sitemap, search and structured-data withdrawal.

23. Consent and relationship disclosures

Identify the project data owner, publication permission scope, whether the company supplied equipment, performed the intervention, received compensation, or has a dealer or sponsorship relationship. A customer quote, photograph, facility name, logo or measured result requires appropriate permission; anonymisation does not automatically authorise disclosure. Record dates, approvals and revocation handling.

24. Publication narrative and visual evidence design

A verified case should present: site context without identifying details unless approved; equipment and metering boundary; baseline demand and dispatch; intervention; matched after demand and dispatch; pressure and process acceptance; measured/adjusted outcome; limitations; persistence; source/consent status. Show before and after traces on comparable axes with time zone, units, missing-data markers and legend. Give each chart a text summary for accessibility and AI extraction.

Keep negative outcomes and exceptions adjacent to headline metrics. A 12% lower average station kW is not a 12% lower annual bill. A reduction in compressor starts may improve an operational metric without demonstrating component-life extension. Do not extrapolate a machine’s rated power to station energy or claim savings from an isolated PLC screenshot.

25. Standards and source applicability

Potential reference families include ISO 11011 for compressed-air energy assessment, ISO 1217 for compressor performance declarations where applicable, ISO 50015 and recognised measurement-and-verification methods for energy outcomes, and relevant electrical metering standards. These are reference candidates only, not assertions of compliance. Record the actual edition, scope, test boundary, calibrated source and reviewer approval before citing any clause or claiming certification. OEM manuals and site process limits control equipment-specific operating constraints.