What the test measures, and what it does not
A compressor’s rated capacity is its free air delivery: the volume of air it takes in at atmospheric conditions per minute, expressed at a stated reference. A pump-up test estimates that figure from first principles. Air is compressed into a fixed volume; the pressure rise tells you how much atmospheric air went in; the time tells you the rate. No flow meter, no laboratory.
What it does not measure is the certified capacity under the test standard, because the standard controls temperature, pressure basis, instrument accuracy and corrections that a site test cannot. It does not measure the air reaching the far end of the plant either, which is a distribution question. And on a screw compressor with modern controls it measures the machine’s behaviour under its control logic, not its airend alone. Used for what it is, a repeatable condition check, it is one of the most useful tools a plant has.
The arithmetic
Under the simplifying assumptions of constant temperature, a fixed volume and no leakage, the free air added to a receiver is the receiver volume multiplied by the rise in absolute pressure, divided by the reference absolute pressure:
free air delivered (m³/min) = V × (P2 − P1) ÷ (P_ref × t)
where V is the volume in cubic metres (receiver plus any connected piping that is pressurised with it), P1 and P2 are the start and finish pressures in bar absolute, P_ref is the reference atmospheric pressure in bar absolute (1.013 bar at sea level; lower at altitude), and t is the time in minutes. Because the formula uses a pressure difference, the result is the same whether you use gauge or absolute pressures for P1 and P2, as long as both are the same kind; the reference pressure must be absolute.
Three things follow. First, the volume must include everything that fills during the test: receiver, the pipe to the first closed valve, any dryer or filter housings in that path. Second, the time must be measured while the compressor is loaded and delivering, so a screw machine’s unload period does not count. Third, the result is free air at the reference pressure, which is what the nameplate quotes, so the two can be compared directly, with the reservations below.
Worked example 1: A piston compressor on its receiver
A two-stage piston compressor sits on a 500-litre receiver. With the outlet valve closed, the receiver is drained to 4 bar gauge, the compressor is started, and the pressure reaches 7 bar gauge in 90 seconds. Volume 0.5 m³; pressure rise 3 bar; reference 1.013 bar absolute; time 1.5 minutes. Free air delivered ≈ 0.5 × 3 ÷ (1.013 × 1.5) = 0.99 m³/min, about 35 cfm or 16.5 L/s. If the nameplate says 1.1 m³/min at 7 bar, the machine is delivering about 90% of its rating, which for a piston machine of some age is a prompt to look at valves and rings, not an emergency.
Worked example 2: A screw compressor with pipework in the test volume
A 37 kW screw compressor feeds a 1,000-litre receiver through a dryer and 30 metres of 50 mm pipe to a closed isolation valve. The pipe holds about 60 litres and the dryer and filters perhaps 40; the test volume is 1.1 m³. The machine is run loaded from 5.5 to 7.5 bar gauge and takes 32 seconds. Free air ≈ 1.1 × 2 ÷ (1.013 × 0.533) = 4.07 m³/min, about 144 cfm. A 37 kW machine of this class is typically rated around 5.5–6.5 m³/min at 7 bar, so this reading says something is wrong: a blocked intake filter, an inlet valve not opening fully, a leaking minimum-pressure valve, or a large leak in the test volume. The next step is a pump-down leak test, below, before anyone blames the airend.
The pump-down test: Measuring leakage first
With the compressor stopped and the test volume isolated from the plant, time how long the pressure takes to fall from one value to another. The same formula gives the leakage rate as free air: leakage (m³/min) = V × (P2 − P3) ÷ (P_ref × t). If the 1.1 m³ volume in the second example falls from 7.5 to 7.0 bar gauge in 4 minutes with the machine off, the leakage is 1.1 × 0.5 ÷ (1.013 × 4) = 0.136 m³/min, nearly 5 cfm, which during the pump-up test was quietly subtracting from the measured delivery. Add it back and the compressor’s real output is closer to 4.2 m³/min; still low, so the fault is in the machine, but the leak is worth fixing in its own right. Leakage measured this way on the whole plant, with everything isolated at the compressor, is also the standard way to put a number on a leak problem before a survey.
Step by step: Running the test safely
- Isolate. Close the outlet valve so the test volume is the receiver and the pipe up to the valve. Confirm the receiver’s working pressure and the safety valve setting; the test must stay well below both.
- Record the volume. Receiver nameplate volume, plus pipe volume (bore area × length) and any housings in the path. Write it down with the date.
- Note the conditions. Ambient temperature, atmospheric pressure or altitude, and whether the compressor is cold or has run for 20 minutes. A warm machine gives a steadier result.
- Drain and set the start point. Drain condensate. Let the pressure settle at the start value with the compressor stopped.
- Start and time. Start the compressor; start the watch when the gauge passes the start pressure; stop it at the finish pressure. On a screw machine, make sure the machine is loaded throughout: set the test band below its unload pressure, or note any unload time and exclude it.
- Repeat. Do it twice more. If the three times differ by more than a few percent, something is changing: temperature, a leak, or the controls.
- Pump down. Stop the compressor, time the pressure fall over a defined band, and compute leakage.
- Correct and compare. Add the leakage back to the measured delivery; compare the result with the nameplate free air at the same discharge pressure.
Never raise pressure to test a safety valve, never close a valve between a vessel and its relief device, and never run a test on a receiver without a current certificate; the safety, isolation and permit-to-work guide sets out the rules.
Where the errors come from
| Source of error | Direction | Size | What to do |
|---|---|---|---|
| Air temperature rising during the test (compression heats the air; the receiver cools it over time) | Overstates delivery | A few percent on a short test; more on a long one | Keep the test short; let the machine warm up first; use the same band each time so results are comparable |
| Leakage from the test volume | Understates delivery | Anything from 1% to 20% | Run the pump-down test and add the leakage back |
| Volume miscounted (pipe, dryer, filters ignored) | Understates delivery | Often 5–10% on small receivers with long pipe runs | Include everything between the compressor and the closed valve |
| Unload or modulation time counted (screw machines) | Understates delivery | Can halve the result | Test below the unload pressure; watch the load indicator |
| Gauge error | Either | A cheap gauge can be 5% out | Use a calibrated gauge or a digital reference; the same gauge for all tests |
| Wrong reference pressure (altitude) | Overstates delivery at altitude | About 1% per 100 m above sea level | Use the site’s atmospheric pressure in the reference |
| Compressor not warmed up | Either | Cold oil and tight clearances change the first minutes | Run 15–20 minutes before testing |
The direction matters. A low result with a leak-free, correctly counted volume is a real finding; a low result with an unchecked leak is a guess.
Reading the result
The useful comparison is the measured free air, corrected for leakage, against the nameplate free air at the same discharge pressure. Site tests cannot match the standard’s accuracy, so a result within about 5% of the nameplate is “as rated”; 85–95% is “worn or restricted, investigate”; below 85% is “a fault, find it”. For a piston machine the usual culprits are valve plates, rings and a choked intake filter; for a screw machine, the intake filter, the inlet valve, the minimum-pressure valve and the separator, in that order, before the airend. A result far above the nameplate means the volume is overstated or the finish pressure was read early.
Just as valuable is the trend. A machine tested at commissioning and then at each major service shows its decline before the plant feels it; the commissioning and acceptance page builds the first reading into the handover record for that reason.
Indian site conditions and the test
At 40–45 °C ambient the air entering the compressor is less dense, so free air delivery falls a few percent compared with the catalogue’s 20 °C figure; this is physics, not a fault, and the manufacturer’s derating curve says how much. Receivers in the sun warm the air inside and bias a long test upward; test in the morning or shade the vessel. At altitude, Hyderabad’s roughly 500 m, atmospheric pressure is about 0.95 bar, so use that as the reference; using 1.013 overstates the result by about 6%. And condensate: a receiver that has not been drained has less volume than its nameplate says, which overstates delivery.
When a pump-up test is not enough
For a contractual acceptance, a warranty claim, an energy audit or a station of several large machines, the pump-up estimate gives way to an instrumented flow measurement on a stated basis: a flow meter of suitable type with its calibration, pressure and temperature at the measurement point, and the correction method recorded. Screw compressors with integrated receivers, VSD machines whose speed follows pressure, and any machine whose controller will not stay loaded through the band are poor candidates for pump-up testing at all. The compressor performance acceptance worksheet sets out the instrumented method and the FAD and reference-condition conversion tool converts between bases.
The interactive calculator needs JavaScript. The worked example below shows the method with fixed numbers.
How Reddy Compressors helps
Our service engineers run pump-up and pump-down tests as part of every major service and commissioning, record the results against the nameplate, and carry out instrumented flow tests where a contract or a warranty needs them. Independent of any manufacturer; more than 20 service engineers and technicians across Telangana and Andhra Pradesh; since 2002.
What to send us
- A photo of the compressor nameplate and the receiver nameplate
- Your test readings: volume, start and finish pressure, time, ambient temperature
- What the plant is seeing: low pressure, long running hours, a machine that never unloads
- WhatsApp 99590 60718
Frequently asked questions
Can I pump-up test a screw compressor?
Yes, as a condition check, provided the machine stays loaded through the test band and you include the pipe and dryer volume. Treat the result as an indicator; for a contractual figure use an instrumented flow test.
Watch
Short explainers on this topic
Slide-by-slide engineering explainers from Reddy Compressors. More in the video library.
Why is my result higher than the nameplate?
The volume is probably overstated, the finish pressure was read early, or the air warmed during a long test. A result more than a few percent above the rating is not a bonus; recheck the inputs.
How accurate is the test?
With a calibrated gauge, a correct volume, a short test and the leakage corrected, within about 5% of an instrumented test for a piston machine. Without those precautions it can be 20% out either way.
Do I need to isolate the whole plant?
Only the test volume: close the valve after the receiver. For a plant leakage measurement, isolate at the compressor and time the fall of the whole network.
What is a normal leakage figure?
Well-maintained plants run at 5–10% of compressor capacity; 20–30% is common in plants that have never been surveyed. The pump-down test gives you the number; the leakage cost scenario calculator puts a price on it.
How often should I test?
At commissioning, at every major service, and whenever the plant reports low pressure or long running hours. The trend is worth more than any single reading.
Related pages
Technical content reviewed by D. Srinivas Reddy, Managing Partner and service engineer, Reddy Compressors. Last updated 4 November 2026. How our engineering content is reviewed.