CAGI Data Sheets Explained How to Compare Compressor Performance Objectively

Compressor sales conversations follow a pattern. The salesperson talks about airend technology, premium efficiency motors, total cost of ownership. A quotation arrives. The buyer compares price, glances at horsepower and cfm, picks the cheaper one. Somewhere in the documentation package sits a CAGI data sheet that stayed in the envelope.

CAGI is the Compressed Air and Gas Institute, a U.S. trade association that runs a voluntary performance verification program. Manufacturers submit data, an independent lab tests randomly selected units, and if the results fall outside tolerance bands the manufacturer corrects the data or withdraws. The tolerance band matters and almost never gets discussed: a tested compressor passes while delivering somewhat less air and drawing somewhat more power than claimed, as long as deviations stay inside the window. What gets published is the best version of the truth within a range, neither a lie nor a per-unit guarantee. Airend machining tolerances, motor winding variation, assembly quality all introduce scatter between serial numbers of the same model.

Coverage is uneven. Oil-injected rotary screws in the 5-500 hp range from North American and European manufacturers: well populated. Oil-free machines, centrifugal compressors, Asian export units: thin. Within a single product family, a manufacturer might list the 75 hp and 125 hp and skip the 100 hp, which usually marks a model sitting at an awkward airend-to-motor match point. Manufacturers volunteer their strong data.

Most comparisons go wrong here.

CAGI inlet conditions are 14.5 psia, 100°F, 100% relative humidity. ISO 1217 Annex C conditions are 1 bar (14.504 psia), 20°C, 0% relative humidity. The same compressor tested at Annex C shows 5-10% higher capacity. Same machine, different yardstick. European manufacturers have historically published Annex C numbers. CAGI participants publish CAGI numbers.

Correcting for this requires a psychrometric density calculation accounting for temperature, pressure, and moisture content differences. It takes about fifteen minutes. It almost never gets done, because the buyer does not know it needs doing.

The reason this matters beyond a percentage on a datasheet: a 200 hp compressor consumes around $50,000 in electricity per year at typical industrial rates. That machine will run for twelve to fifteen years. If the capacity comparison is off by 7% because of an inlet condition mismatch, the resulting mis-sizing cascades through the entire operating cost profile. Oversizing puts the machine in unloaded or modulation mode for excessive hours, consuming 25-40% of full-load power while delivering zero air. Undersizing forces the compressor to run at full load with no margin and creates pressure instability that triggers false alarms and production interruptions downstream. Both outcomes trace back to a capacity number that was never properly compared.

European manufacturers selling into North America benefit from the Annex C inflation and have limited incentive to explain the discrepancy. The CAGI-rated competitor can point it out, but doing so effectively requires the buyer to understand the issue, and most sales engineers find it easier to compete on price or service terms than to deliver a tutorial on psychrometrics during a sales call. So the confusion persists.

The 100% relative humidity in the CAGI spec adds yet another layer. Saturated inlet air means the compressor ingests air already at maximum moisture load. Water vapor displaces dry air volume. When CAGI reports capacity in acfm, the figure accounts for this moisture displacement. Some competing standards do not handle moisture the same way, adding another 1-2% of apparent divergence on top of the temperature and pressure effects.

Specific power is only comparable at the same rated pressure. Each 2 psi increase in discharge pressure costs about 1% more input power. Some manufacturers publish CAGI data at 100 psig even though most of their installed base runs at 125. CAGI does not mandate a rating pressure.

Total package input power is measured in kilowatts at the input terminals with a power analyzer. Non-CAGI specifications often quote shaft power or nameplate hp, both lower. On a 100 hp air-cooled rotary screw, the cooling fan alone pulls around 5 hp. Two identical airends with identical motors show different CAGI total package power if one has a variable-speed fan drive and the other a fixed-speed axial fan. A few CAGI sheets break out fan power as a separate line. When that line exists, compare it.

Specific power is total package input power divided by capacity. kW per 100 cfm. For the 100-200 hp class at 125 psig, full-load numbers cluster around 19 kW/100 cfm, best near 18, worst near 22. Compressed air is routinely the most expensive utility in a manufacturing plant on a per-unit-of-energy basis, and specific power is the metric that ties the compressor purchase to that ongoing cost.

The CAGI figure is taken with a new oil separator element, a clean inlet filter, and fresh oil. In the field, all three accumulate contamination from the moment the compressor starts. The gap between the published number and the operating number over the other 99.9% of the machine's life is, for most installations, larger than the gap between competing manufacturers' published numbers.

The separator element is a coalescing filter that strips entrained oil from the discharge air-oil mixture. New element pressure drop runs about 4 psi. Over thousands of hours, oil residue and fine particulate accumulate and the drop climbs. By mid-life it reaches 8 psi. In applications with poor oil maintenance or high ambient contamination it goes into double digits. The airend sees higher backpressure and works harder. A machine that tested at 18.5 kW/100 cfm with a new separator is running north of 19.5 by the time the element is due for replacement.

Manufacturers who specify 8,000-hour separator replacement intervals instead of 4,000 are saving their customers money on parts and costing them more on electricity. Parts cost appears on a maintenance invoice that someone reviews and approves. Electricity cost is allocated across the entire facility. Guess which one gets optimized.

Inlet filter loading works the same way from the intake side: starves the airend of mass flow while power holds roughly constant. In dusty environments the capacity penalty reaches 3% or more before anyone notices.

Plant managers track bearing vibration, oil particulate count, and discharge temperature to three decimal places, but somehow do not track the one measurement that would tell them whether the compressor is still performing as quoted. Electricity is a fixed overhead in most plant accounting systems, not a variable cost attributed to specific equipment, so there is no line item that anyone is responsible for optimizing. The CAGI data sheet sits in a filing cabinet in the maintenance office. The utility bill goes to accounts payable in a different building.

In compressed natural gas (CNG) fueling station applications, where compression cost is the primary revenue driver, operators monitor specific power obsessively and replace separator elements at the first sign of elevated pressure drop. The incentive structure is different. In industrial compressed air, the incentive to monitor exists but the organizational structure to act on it usually does not. The compressor report lands on the maintenance manager's desk. The electricity bill lands on the facilities manager's desk. In many plants, those are different people with different budgets.

Manufacturers build a set of airend sizes and pair each with a range of motors. One casting shows up at 75, 100, and 125 hp with different drive speeds. At one of those pairings the rotor tip speed, the built-in volume ratio, and the rated pressure ratio all align and specific power hits its minimum. At the other pairings the airend is either over-sped or under-sped and specific power suffers. This explains something that confuses buyers: why the same manufacturer's 100 hp compressor sometimes has worse specific power than their 75 hp. It is not that the bigger machine is poorly built. It is sitting at a point in the product line where the available airend is slightly too large for the motor or vice versa. The compromise shows up in the CAGI number. Knowing this, a buyer needing 400 cfm who sees the 75 hp at 420 cfm and 18.2 kW/100 cfm, and the 100 hp at 530 cfm and 19.4, should consider the smaller machine loaded near its sweet spot, provided it has enough capacity margin to cover demand peaks.

Two-stage compression gets oversold below 125 psig. At 150 psig and above, splitting the pressure ratio between two stages with intercooling produces a clear thermodynamic benefit even after accounting for the added mechanical complexity. Below 125, a good single-stage airend with tight clearances and favorable L/D ratio will match or beat a mediocre two-stage on specific power.

CAGI publishes a second sheet for VSD compressors with specific power at 40, 60, 70, 80, and 100 percent of full-load flow. At full load the spread between manufacturers is narrow. At 40% it gets wide. A VSD compressor designed from scratch for variable-speed duty holds fairly flat specific power across the speed range. An afterthought VSD, where the inverter is an add-on to an existing fixed-speed design, performs about the same at full speed but deteriorates below 70%, because the airend's built-in volume ratio becomes mismatched and internal leakage (roughly constant regardless of rotor speed) becomes a larger share of the reduced throughput. For a plant that spends most of its production hours between 50% and 70% of peak demand, the part-load profile sets the annual electricity cost and the full-load number is beside the point.

Confirm both units carry CAGI verification. Verify rated pressure is the same on both sheets. Look at inlet condition basis: two CAGI sheets need no correction, a CAGI sheet next to an Annex C claim needs a density correction before anything else. Match the comparison to the demand profile, weighting part-load data by hours spent at each level for VSD trim machines. If cooling methods differ, model downstream dryer energy. Note the drive type: belt drives lose a few percent that creeps up as belts stretch, gear drives cost a couple percent at the mesh, direct coupling loses nothing. At altitude, correct for air density.

The DOE's Compressed Air Challenge has documented leak rates around 25% of total air production as typical across hundreds of facility assessments. Pressure drops of 15 psi or more between compressor discharge and point of use are common in aging distribution systems. A compressor with 18 kW/100 cfm specific power installed in a system leaking a quarter of its output delivers an installed specific power above 24. The piping, the fittings, the abandoned drops, the quick-connect couplings that have been leaking for years, the open blowoffs that someone rigged for cooling. That is where the money goes.

Published full-load specific power across major manufacturers has converged into a band of about 5-8%. Rotor profiles are no longer a competitive differentiator. Motor efficiencies have standardized at IE3 and IE4. Meanwhile the spread in VSD part-load performance, in multi-compressor system controllability, and in how fast efficiency degrades between maintenance intervals remains much wider. Full-load specific power absorbs the most comparison effort while offering the least remaining differentiation.

CAGI data sheets say nothing about bearing life or oil carryover concentration. They do not cover vibration, controller protocols, or how easy it is to change the oil filter without removing three panels. Spare parts pricing is absent. Service network coverage is absent. A separate CAGI sheet covers sound levels. Reliability and maintenance cost have no standardized reporting format in this industry.