Compressed Air: Buy the Required Service, Not the Largest Compressor

Compressed Air & UtilitiesBaiyX ResearchMay 2026Updated November 2026

Download report text · Download exhibit data (CSV)

The investment conclusion

The best compressed-air investment is often a better-defined requirement before it is a new compressor. A plant may be paying to generate air that leaks away, operating at excessive pressure to compensate for a local restriction, or keeping several machines running inefficiently because controls do not coordinate them. Replacing the oldest compressor can help, but it can also preserve the underlying waste in a newer package. Our recommendation is to measure the system, reduce avoidable demand, establish the required pressure and quality, and only then size the supply equipment and evaluate competing quotations.

This sequence is commercially important because compressed air is an on-site utility. Its value comes from enabling production at the required reliability and quality, not from maximising compressor output. A lower purchase price can be overwhelmed by electricity and maintenance over a long operating life. A high-efficiency machine can also disappoint if it operates outside its favourable range, supports unnecessary pressure or cannot be serviced promptly. The relevant comparison is the cost of delivering usable air to production, including treatment, distribution, controls and recovery arrangements. That boundary is wider than the compressor skid shown in a catalogue.

For BaiyX, this creates a connected equipment and aftermarket proposition: compressors, treatment equipment, traceable service parts and maintenance consumables can be assessed as one operating system. The opportunity is strongest when the supplier can document compatibility, explain performance conditions and help the customer make a measurable improvement. It is weaker when the offer consists only of a lower price on a nominally equivalent machine. This report therefore concentrates on the decisions that change plant economics and on the evidence a buyer should request before treating an efficiency claim as a bankable saving.

Use historical technical guidance honestly

The US Department of Energy’s August 2004 compressed-air cost tip sheet explains why operating hours, loading and electricity price matter and promotes a systems approach. Its familiar illustrative lifecycle split assigns 76% to electricity and 12% each to equipment and maintenance. That chart is a historical example, not a universal current cost structure. We do not use its electricity tariff or its cost per volume of air as a 2026 benchmark. A modern project needs its own measured power, operating pattern, local tariff, maintenance scope and installed investment.

Evidence: US Department of Energy, Determine the Cost of Compressed Air for Your Plant, Tip Sheet 1, August 2004; historical technical reference

The distinction between durable engineering guidance and current commercial evidence is essential. The underlying need to measure power across operating states remains useful. A decades-old price input does not become current because it appears in an authoritative publication. Similarly, a supplier’s statement that energy dominates lifetime cost may be directionally sensible but insufficient for procurement. A standby compressor used occasionally has a different cost structure from a continuously loaded production unit. A plant with low electricity prices and expensive service access faces a different trade-off from a plant with high tariffs and strong local support.

We therefore separate three kinds of evidence throughout this report. Published technical references explain mechanisms and measurement approaches. Verified supplier data describe equipment under specified test conditions. The financial examples use explicitly hypothetical inputs to show how a buyer can calculate its own result. None should be mistaken for a market-wide promise. This separation improves the commercial discussion: suppliers can compete on the equipment evidence they control, while the buyer remains responsible for the operating assumptions and production consequences that only its own plant can establish.

Define usable air at the point of demand

A compressed-air requirement has at least four dimensions: flow, pressure, quality and availability. Flow must be defined with its reference conditions, because volumetric figures can refer to different pressure and temperature bases. Pressure must be measured where it matters to production, not only at the compressor discharge. Quality must address the contaminants and moisture relevant to the process. Availability must reflect the consequence of interruption. A quotation that specifies only motor kilowatts and a nominal flow leaves too much room for suppliers to offer materially different services under superficially similar descriptions.

The buyer should map the main users and identify the most demanding legitimate requirement. Some equipment needs a short high-flow pulse; other equipment draws a steady amount. Some processes need especially dry or clean air, while ordinary pneumatic tools may have different requirements. It can be wasteful to impose the most demanding specification on the entire plant if a local treatment or pressure solution is more economical. It can also be unsafe or damaging to relax a requirement without understanding the process. The task is to separate genuine production needs from inherited settings and undocumented preferences.

Prepare a utility specification with measured minimum and maximum conditions, the expected demand profile and the consequences of noncompliance. Include ambient temperature, ventilation, altitude where relevant, cooling arrangements and the available electrical supply. These conditions can affect equipment selection and usable capacity. Ask suppliers to state their design assumptions in the same document. When one quotation includes sufficient drying and another does not, the apparent price difference is a scope difference. The buyer should resolve that difference before comparing efficiency or negotiating the final equipment discount.

Measure before selecting capacity

A demand survey should capture representative production, including shift changes, breaks, weekends, cleaning and unusual high-demand operations. A brief reading taken during a busy hour cannot establish annual energy use or the appropriate control strategy. The survey should record package electrical power, relevant pressures and flow where suitable instrumentation is available. Time alignment matters: a pressure dip is more informative when it can be connected to a specific demand event and compressor response. The measurement plan should be designed by competent personnel and installed without compromising the safety or integrity of the system.

The resulting data should distinguish productive demand from background demand. Air consumed when production is stopped can indicate leaks, uncontrolled uses or equipment left pressurised unnecessarily. It is not automatically all leakage. Some instruments, purges or essential services may legitimately remain active. The plant should identify those uses before estimating savings. Similarly, a compressor running for many hours is not necessarily delivering much useful air. Unloaded operation and inefficient part-load behaviour can consume electricity while contributing little flow, which is why operating hours alone are an inadequate basis for the investment model.

A useful first deliverable is a demand-duration profile: how many hours the plant spends in each broad flow range, together with the pressure and quality required in those periods. This profile is more informative than a single peak. Capacity must accommodate credible peaks, but energy cost is accumulated across the whole year. The buyer should also identify whether the peak is sustained, intermittent or avoidable through sequencing. A short event may be addressed by storage or local changes rather than a permanently oversized compressor. The right answer depends on the measured system, not on a general preference for more reserve capacity.

Pressure is a service requirement and a cost variable

A plant often raises compressor pressure when a remote machine experiences a low-pressure alarm. That response can mask a distribution restriction, a clogged filter, an undersized local connection or a short demand event. The compressor then works harder to solve a problem that may be local. The better diagnostic sequence is to measure pressure at the compressor, after treatment, across key distribution sections and at the affected user during the event. This reveals where pressure is being lost and whether the remedy should be supply, storage, controls or distribution maintenance.

DOE’s systems sourcebook gives a rule of thumb near 100 psig: reducing discharge pressure by 2 psi can reduce full-output energy use by approximately 1%, with additional effects where unregulated demand changes. This is a screening relationship under stated conditions, not a universal guaranteed saving. Actual results depend on compressor type, control behaviour, demand and pressure requirements. Buyers should use performance curves and measured system data for final calculations, and any pressure change must preserve the requirements of the equipment and processes being supplied.

Evidence: US Department of Energy, Improving Compressed Air System Performance, third edition, 2016; technical reference

The economic value of pressure reduction is therefore conditional. If the plant can remove a restriction and maintain the required point-of-use pressure at a lower header setting, it may reduce both compressor work and some unnecessary demand. If lowering the setting causes production faults, the project has failed regardless of the apparent electricity saving. Define the acceptance boundary accordingly: minimum pressure at the critical users during representative production, together with measured system power. A proposal that promises lower pressure without proving usable service is not an efficiency improvement; it is an incomplete operating change.

Leaks require a programme, not a one-time campaign

Leak repair is attractive because it can reduce demand without changing the productive process. But finding a leak and recording a theoretical volume is not the same as realising an electricity saving. The compressor controls must respond to the reduced demand. If the machines remain running in an inefficient state, some expected saving may not appear at the meter. A leak programme should therefore combine detection, prioritised repair and post-repair measurement, followed by control adjustments where appropriate. The final result is a change in system energy while preserving production, not a count of tagged leaks.

Prioritise repairs by consequence and practicality. A large accessible leak on a noncritical branch may be an immediate opportunity. A small leak requiring a major shutdown may need to be scheduled with other work. Some leaks can also indicate deteriorating hoses, connections or equipment that deserve broader attention. Record the location, estimated severity, repair owner, completion date and verification. Without an owner and follow-through, a survey becomes an inventory of known waste. Without repeat surveys, the system can gradually return to its previous condition as new leaks develop and temporary repairs deteriorate.

Avoid claiming all reductions as permanent annual savings from day one. Production patterns may change, repairs may fail and new equipment may be added. Establish a baseline adjusted for the principal drivers of demand, then review consumption periodically. A simple plant-level metric can be useful if its limitations are understood: energy per unit of production may improve because product mix changes, not because the air system becomes more efficient. Retain enough operating context to explain the result. The objective is a maintenance practice that sustains lower waste, rather than a short-lived improvement selected for a project presentation.

Compressor selection follows the demand profile

Different compressor technologies and control arrangements suit different duties. The buyer should avoid deciding that one technology is always best based on a generic efficiency claim. What matters is how the proposed package performs across the plant’s required flow and pressure range, how it responds to variation and how it can be maintained locally. Ask suppliers to map their proposal against the measured demand-duration profile. A technically credible proposal should explain which unit carries base demand, which unit handles variation and what happens when a machine is unavailable.

Variable-speed control can be valuable when demand varies within a suitable operating range. It is not a substitute for correct sizing or coordinated sequencing. A variable-speed machine operating continually near an unfavourable limit may offer less benefit than expected. A fixed-speed machine carrying stable base demand may be economically appropriate. The comparison should use package power at relevant load points, including auxiliary consumption and unloaded behaviour where applicable. The nominal motor rating is not the same as measured electrical input to the complete package, and maximum output is not the same as annual useful production.

Redundancy also changes selection. A plant that cannot tolerate an interruption may need a standby arrangement or the ability to meet critical demand with one unit out of service. This reserve has capital and maintenance costs even when it consumes little electricity. Define whether redundancy covers all production or only essential loads. A carefully segmented system may preserve critical operations with less installed reserve than a design that treats every air user as equally important. The best capacity plan is one that meets the agreed service and recovery requirements without using oversized equipment as a substitute for understanding them.

Verified data improve comparison, but do not replace site evidence

CAGI’s Performance Verification programme provides a framework for independent verification of published rotary compressor and refrigerated dryer performance within its stated scope. Its current page describes compressor participation from 5 to 200 horsepower and refrigerated dryers from 50 to 1,000 scfm. Standardised datasheets help compare package power, flow and related conditions. Buyers should check the exact model, applicable programme status and test basis; a company’s participation does not mean every product it sells is independently verified or that a datasheet alone proves installed system performance.

Evidence: CAGI, Performance Verification program and datasheet framework; accessed 9 November 2026

The procurement use is practical. Request the relevant datasheet for each proposed operating pressure and configuration. Confirm that units and reference conditions are comparable. Ask whether accessories included in one proposal but excluded in another affect the power boundary. If a supplier provides a self-declared sheet outside a verification programme, identify it as such rather than treating the formatting as proof of independent testing. An alternative test report can still be useful if the method, equipment, uncertainty and issuing body are clear. The objective is consistent evidence, not a preference for a logo detached from its scope.

Laboratory data answer how the package performs under specified conditions. The site question is how the complete installation behaves over its actual duty. Inlet conditions, treatment pressure drop, control settings, maintenance condition and demand variation can change the outcome. The buyer should therefore use verified data to shortlist and model, then use commissioning measurements to confirm the installed result. This two-stage approach is fair to suppliers: it separates equipment performance from site factors while still requiring the complete project to deliver the service and economics on which the purchase was approved.

Air treatment is part of the production process

Dryers, filters, drains and separators are sometimes treated as accessories to be reduced when the compressor quotation exceeds budget. That can be a false economy if moisture or contamination affects products, tools or controls. The correct treatment requirement depends on the application. A general industrial air system and a sensitive process do not necessarily need the same arrangement. The buyer should define the required quality at the relevant point of use, then ask the supplier to explain how the proposed treatment achieves it under the worst credible operating conditions, not only at a favourable rated inlet condition.

Treatment has both direct and indirect costs. Equipment consumes energy or purge air in some configurations, imposes pressure drop and requires maintenance. Overspecifying treatment can waste money; underspecifying it can create quality failures whose cost dwarfs the compressor saving. The analysis should include the full train and its operating conditions. Ask what happens when filters load, drains fail or the dryer is bypassed during maintenance. If a bypass permits production to continue with unacceptable air quality, that arrangement needs explicit operating controls and a clear understanding of the consequence.

Commissioning should verify the agreed quality using an appropriate method and sampling location. A dryer’s status light is not the same as evidence at the production user. Monitoring frequency should reflect consequence: a critical process may need ongoing measurement, while a less sensitive application may use periodic checks and maintenance indicators. Document the limits, response actions and responsibility for deviations. This turns air quality from an assumption hidden inside the equipment specification into an operating requirement that production and maintenance can manage throughout the system’s life.

Distribution and storage can change the investment case

The distribution network determines whether air generated at the compressor reaches users at useful pressure. Long runs, restrictive fittings, poorly maintained treatment equipment and unsuitable branch arrangements can create local problems. A buyer considering a larger compressor should first determine whether the existing supply is genuinely insufficient or whether distribution prevents it from serving the demand. The distinction can change both capital expenditure and ongoing energy cost. A larger compressor may raise the available flow while leaving the underlying restriction and pressure penalty in place.

Storage can help manage short demand events and support control stability when properly designed. Its value depends on the usable pressure range, demand duration and the way it connects to the system. A receiver is not an unlimited substitute for compressor capacity, and pressure-vessel design and installation require competent engineering and applicable local compliance. The procurement question is whether a proposed storage arrangement addresses an observed event with a documented calculation and safe design. A generic recommendation to add a large tank does not establish that the critical user will receive the required pressure when the event occurs.

Distribution projects should be evaluated at the system boundary. If a local improvement allows lower compressor pressure, include the resulting measured energy change. If it simply improves production reliability without reducing electricity, that can still be a valuable result. Do not force every project into an energy-saving narrative. Avoided process faults, lower scrap and reduced maintenance may be the primary benefits. The financial model should state those benefits separately, with evidence and assumptions appropriate to each. This keeps the engineering solution aligned with the actual problem rather than with the type of saving that is easiest to advertise.

A five-year package comparison

Consider an explicitly hypothetical plant operating 6,000 hours annually. Proposal A requires 90 kW of average package input over the agreed duty, while Proposal B requires 78 kW. Installed investment is $55,000 for A and $75,000 for B. Annual maintenance is $4,000 and $4,500 respectively. Electricity costs $0.12 per kWh. These are invented inputs for a reproducible model, not quotations or typical industry values. Both proposals are assumed to deliver the same required usable air and reliability; if they do not, an energy-only comparison is invalid.

Annual electricity use is 540,000 kWh for A and 468,000 kWh for B. Annual electricity cost is therefore $64,800 and $56,160. Adding maintenance gives annual operating cost of $68,800 for A and $60,660 for B. B saves $8,140 annually after its $500 maintenance premium. Its additional $20,000 investment has a simple incremental payback of approximately 2.46 years. Over five undiscounted years, total cost is $399,000 for A and $378,300 for B, a difference of $20,700 in favour of B.

At a 10% discount rate, the five-year annuity factor is approximately 3.7908. Present-value cost is about $315,807 for A and $304,943 for B, so B’s discounted advantage is approximately $10,863. The calculation excludes tax, finance structure, residual value and demand charges, which the buyer should add if material. The example demonstrates why a moderate efficiency improvement can justify a capital premium at sufficiently high hours and tariffs. It does not establish that a more expensive compressor is always more efficient or that supplier estimates of average input power should be accepted without evidence.

Electricity price and utilisation determine the premium

The example’s annual energy saving is 12 kW multiplied by 6,000 hours, or 72,000 kWh. At $0.06 per kWh, gross energy saving is $4,320 and net saving after additional maintenance is $3,820. At $0.18, gross saving is $12,960 and net saving is $12,460. Simple payback on the $20,000 premium changes from about 5.24 years to 1.61 years. This range is a sensitivity analysis, not a forecast of future tariffs. It shows why a supplier’s payback claim cannot be transferred from one country or factory to another without rebuilding the inputs.

The five-year discounted break-even tariff can be calculated directly. The premium requires annual net saving of $20,000 divided by 3.7908, approximately $5,276. Add the $500 maintenance premium and divide by 72,000 kWh: the break-even electricity price is approximately $0.0802 per kWh. Above that price, B has the lower present-value cost under the stated assumptions. Below it, A does. If operating hours fall to 3,000 while the power difference remains the same, the energy saving halves and the break-even tariff roughly doubles. This is why standby duty and continuous duty should not be judged by the same purchasing rule.

The model is only as good as the assumed 12 kW difference. If that difference exists at full load but disappears across the real demand profile, the calculated saving is overstated. Ask the supplier to show the weighted power calculation across measured operating states. Include unloading, standby and treatment energy where they fall inside the project boundary. If demand changes materially after leak repair or production expansion, update the model before ordering. The most valuable analysis is often the one that prevents the plant from buying equipment for a demand profile that no longer exists.

Avoid double counting improvement measures

Compressed-air projects often combine leak repair, pressure reduction, control changes and new equipment. Each measure can be valuable, but their savings are not necessarily additive. Repairing leaks reduces the flow against which a new compressor’s efficiency advantage is calculated. Lowering pressure can change both compressor power and unregulated demand. A control improvement can alter the number of running machines. Adding separate percentage savings taken from different studies can therefore produce an implausible total. The model should apply measures sequentially to a consistent baseline or use a system simulation grounded in measured data.

A simple hypothetical illustration makes the issue clear. Start with 500,000 kWh per year. Assume a verified demand project reduces energy by 10%, leaving 450,000 kWh. If a subsequent equipment change reduces the remaining consumption by 12%, it saves 54,000 kWh and leaves 396,000 kWh. Total saving is 104,000 kWh, or 20.8%, rather than 22%. The percentages here are invented for arithmetic demonstration. Real interactions can be more complicated, especially when compressor control behaviour changes, but the basic discipline is the same: apply each improvement to the condition that actually exists after the preceding change.

Project reporting should state the measurement boundary and sequence. If the utility meter includes other equipment, isolate the relevant load or use an appropriate adjustment method. If production changes, explain how the comparison accounts for it. Do not credit a compressor replacement for savings already achieved through a prior shutdown policy. Equally, do not dismiss a sound project because total factory electricity rose while production expanded. The goal is a credible attribution of improvement that can survive review by operations and finance, rather than a persuasive percentage assembled from incompatible baselines.

Maintenance determines whether efficiency persists

A compressor’s performance at commissioning is a starting point. Filters load, coolers foul, drains malfunction and operating conditions change. The maintenance plan should identify which conditions affect energy, capacity, air quality and reliability, and how they are detected. A calendar schedule alone may be insufficient when duty and environment vary substantially. At the same time, condition-based maintenance should not become an excuse to ignore the manufacturer’s requirements or the limits of available measurements. The appropriate plan combines documented service requirements with observations that reveal deterioration before it becomes a production problem.

Consumable compatibility deserves particular attention in cross-border sourcing. A replacement filter, separator element or lubricant should be qualified against the actual equipment and duty. Similar dimensions do not establish equivalent performance. Request the relevant specifications, traceability and instructions, and understand whether a substitution affects warranty or service arrangements. A low-cost consumable can be economically attractive when it is technically suitable and consistently supplied. It can be expensive when it increases pressure drop, shortens service life or introduces uncertainty that the maintenance team cannot diagnose. The purchasing decision should consider installed operating cost, not only the price of the item.

Keep records that connect maintenance to system performance. Note the service date, parts used, operating hours, pressure differential where relevant and any change in power or quality. If energy rises gradually after service intervals lengthen, the plant can investigate rather than relying on memory. If a new consumable performs well, its evidence supports a qualified second source. This creates a useful aftermarket relationship: the supplier helps the customer preserve measurable performance and reduce supply risk. It also gives the buyer a factual basis for challenging both overpriced proprietary supply and poorly documented alternatives.

Reliability should be valued separately from efficiency

A highly efficient air system that stops production unpredictably is not a successful investment. Reliability requirements should therefore be defined before comparing energy cost. Identify critical users, tolerable interruption and the available fallback. Some plants can stop nonessential branches and continue key operations; others have processes that cannot recover easily from a pressure loss. These differences affect redundancy, storage, maintenance windows and spare-parts policy. An efficiency model that assumes uninterrupted operation while excluding the cost of achieving it understates the true investment.

Consider a hypothetical production line losing $2,000 of contribution for each hour of unplanned air interruption, with no recovery through later production. A four-hour event costs $8,000 before repair expense. If a backup arrangement costs $12,000 annually to own and maintain, it would need to prevent at least six such hours per year to break even on this simple basis. That does not mean six hours of failure should be assumed. The buyer must estimate event likelihood and consequence, and may also value protection against a rare severe event that threatens customer commitments or process integrity.

Service capability should be tested at the destination. Confirm who answers the call, what can be diagnosed remotely, where critical parts are stocked and which technicians can work on the proposed equipment. Ask for a realistic restoration sequence rather than a vague response-time promise. Arriving on site is not the same as restoring air. The plan should distinguish diagnosis, part availability, repair, testing and return to production. This level of detail helps the buyer decide whether a lower-cost machine with a stronger support package is preferable to a technically sophisticated package whose local recovery route remains uncertain.

Contract for performance within a defined boundary

A useful purchase specification states usable flow, pressure, quality, operating range and site conditions, together with the power and reliability evidence required. It should identify what is included in the package and what is provided by the buyer. Suppliers should disclose the assumptions used in their annual energy calculation and the limits of any guarantee. A promise to save a percentage of the plant’s bill is difficult to enforce if the baseline, production level and measurement boundary are undefined. A narrower, measurable commitment is more valuable than an impressive claim that depends on conditions no one has agreed.

Acceptance should combine documentation and measurement. Verify configuration, safety and installation requirements through competent personnel, then measure the agreed operating points and system behaviour under representative demand. Confirm that pressure and quality remain adequate at critical users. Record control settings, sequencing, alarms and maintenance information. Where a full annual duty cannot be observed during commissioning, distinguish the measured points from the model used to extrapolate them. The buyer can then monitor actual operation and resolve differences without pretending that a short test proved every future operating condition.

Commercial remedies should be proportionate and technically meaningful. If power exceeds the agreed condition, the supplier may need to correct settings, rectify equipment or provide another agreed remedy. If the buyer changes demand or site conditions, the effect should be assessed transparently. Avoid contracts that make the supplier responsible for all factory energy regardless of scope, or that leave the buyer with no recourse because every result is described as indicative. A balanced agreement links payment and acceptance to evidence within each party’s control, with a clear process for investigating deviations.

When replacement is the wrong answer

The strongest argument against purchasing a new compressor is that the current system may already have enough efficient capacity once avoidable demand and controls are addressed. A leak programme, shutdown policy, distribution improvement or sequencing change can sometimes release capacity at lower capital cost. This is not a universal claim that maintenance always beats replacement. An ageing machine with poor support, unacceptable reliability or unsuitable performance may still need to be retired. The decision should follow a comparison of complete alternatives, including the cost and remaining life of the existing equipment.

Another reason to defer replacement is uncertain demand. If a production line is about to change, buying against the old profile can create a long-lived mismatch. Temporary arrangements or a staged capacity plan may preserve flexibility. Conversely, waiting indefinitely can be expensive when measured energy and maintenance losses are already clear. The buyer should estimate the cost of delay: expected monthly operating disadvantage, failure exposure and the opportunity to coordinate installation with a planned shutdown. This converts a vague debate about whether the machine is old into a time-sensitive capital allocation decision.

Our preferred ranking is to eliminate obvious waste, verify the remaining requirement, optimise the system configuration and then select equipment. Where a critical reliability issue exists, address it immediately while the broader efficiency work proceeds. The objective is not to minimise capital expenditure at all costs. It is to purchase the least expensive dependable service over the relevant horizon. A disciplined buyer may ultimately select a premium compressor, a simpler package, an overhaul or a combination of smaller changes. The correct answer is the one supported by measured duty, realistic support and transparent economics.

Design a measurement plan finance can audit

A useful energy baseline is reproducible. Record the meter or instrument, its location, the time interval, the operating period and the equipment included. Note whether input power is measured directly or estimated from other variables. Direct package power measurements can avoid some uncertainty associated with inferring electrical consumption from motor ratings, but instrumentation still requires suitable installation, calibration and interpretation. The data owner should retain the original readings, not only a chart copied into a supplier presentation. This allows engineering and finance to inspect how the proposed saving was calculated and whether the comparison remains valid after operating changes.

Time resolution should match the question. Annual billing data can establish the scale of electricity expenditure but may hide short cycling and brief pressure events. High-frequency logging can reveal those events but may cover too few production patterns to estimate the full year. A sound study combines an appropriate short-period diagnostic view with a representative operating calendar. Document unusual conditions such as shutdowns, overtime, product trials or temporary rental equipment. The purpose is not to create an unnecessarily elaborate monitoring project. It is to avoid a precise-looking annual number based on a week that does not resemble the year.

Uncertainty should be visible where it changes the decision. If two proposals differ by only a small amount of modelled energy, measurement uncertainty and demand variability may be larger than the apparent advantage. In that case, service, maintainability and capital flexibility may deserve greater weight. If the difference remains large across conservative assumptions, the energy case is more robust. Present a range or sensitivity rather than adding false decimal precision. A buyer should be able to explain which conclusion is stable, which depends on future utilisation and which requires another measurement before capital is committed.

Build a service-parts policy around consequence

Not every maintenance item deserves the same inventory policy. Frequently used consumables can be planned against service intervals and observed consumption. A rare electronic or mechanical component may deserve stock because its absence would stop a critical utility. Other parts may be obtained through a verified local service pool. The decision should consider lead time, failure consequence, interchangeability, storage requirements and obsolescence. A spare held without the skill, tools or configuration required to install it provides less protection than its purchase price suggests. The recovery procedure and the inventory decision should be reviewed together.

For imported equipment, ask the supplier to identify parts that are standard locally and parts that remain dependent on the original manufacturer. Confirm exact part identification and approved alternatives before the first emergency. Where the supplier offers a consumables kit, examine its contents and service assumptions rather than accepting a generic annual package. The correct kit depends on hours, environment and equipment configuration. Traceability matters particularly when substitute components affect pressure drop, lubrication or separation performance. The buyer should be able to connect the supplied item to an approved specification and to the asset on which it is used.

The policy should evolve with the operating record. A component that fails repeatedly may require a design or maintenance correction instead of a larger stockholding. A part that never moves may still be justified as protection against a severe outage, but that rationale should be explicit. Review critical stock when equipment is modified, firmware changes or a model is discontinued. This avoids the familiar situation in which a factory owns a valuable shelf of spares that no longer match its operating equipment. The objective is dependable recovery at reasonable total cost, supported by evidence from the actual installed system.

A practical 90-day programme

During the first month, appoint an owner for compressed-air performance and assemble the baseline. Map the major users, identify critical pressure and quality requirements, review service history and collect representative demand and power data. Record the operating conditions and measurement limitations. Separate immediate maintenance defects from projects requiring engineering review. The deliverable is a defined utility requirement and an initial loss map, not simply a list of compressor model numbers or an estimate based on motor nameplates.

During the second month, implement appropriate low-risk corrective work and remeasure the system. Develop alternative supply configurations using the remaining demand profile. Request comparable package data, maintenance scope and local service evidence from suppliers. Build a lifecycle model with the plant’s tariff and operating hours, and test lower utilisation, higher maintenance and longer recovery scenarios. Review air treatment and distribution at the same time. If the proposal solves one problem by creating another, such as lower energy with inadequate quality or insufficient reserve, revise it before commercial award.

During the third month, approve the preferred operating plan and finalise acceptance, measurement and maintenance arrangements. Record which savings are measured, which are modelled and which remain contingent on future changes. After commissioning, review actual performance against the baseline and assign responsibility for sustaining it. The most valuable output is not a single efficiency percentage. It is a utility system whose cost, quality and recovery behaviour are understood well enough that production can rely on it and management can make the next investment decision with evidence rather than inherited assumptions.