What Specific Speed Tells You About a Pump You Have Never Seen
Why this matters
You can stand in front of an unfamiliar pump, take five numbers off its curve sheet or nameplate, and know within a minute whether starting it against a closed discharge valve will trip the overload or protect the motor. Those are opposite answers, they come from the same machine family, and the number that separates them is specific speed. It is not an efficiency figure and it is not a rating. It is a shape number: it tells you what the impeller looks like inside, and impeller shape decides which way the power curve runs as flow changes. Getting that backwards is how a start-up crew trips a large pump three times and then jumps the overload.
This card is the calculation, the two corrections almost everyone drops, and two filled-in shape cards for real machines at opposite ends of the range.
The number, and the four conditions it is defined under
Specific speed in US customary practice is the shaft speed in rpm, times the square root of the flow in gpm, divided by the head in feet raised to the three-quarter power. It carries four conditions, and every one of them changes the answer:
- At the best efficiency point, not at the rated point, not at the duty the job needs. A pump has one specific speed and it is defined at BEP.
- Head per stage. A multistage pump's total head is divided by the number of stages before the exponent is applied, because the number describes one impeller.
- Flow per suction eye. A double-suction impeller is fed from both sides, so it takes half the total flow before the square root is applied.
- At full impeller diameter, since a trimmed impeller is no longer the shape the number was derived for.
It is also dimensional, which means the value depends on the unit system it was computed in. The metric conventions produce a number roughly fifty times smaller for the same machine, so a value read out of one reference and compared to bands from another is meaningless. Write the units beside the number every time.
What the value predicts
| Approximate value (US customary) | Impeller shape | Head curve | Shaft power as flow falls | Starting against a closed valve |
|---|---|---|---|---|
| 500 to 2,000 | Radial vane, narrow, large diameter | Steeper, wide stable range | Falls, minimum at shutoff | Standard practice, briefly |
| 2,000 to 4,500 | Francis, wider passages | Moderate | Nearly flat with flow | Usually acceptable, confirm |
| 4,500 to 8,000 | Mixed flow | Flatter | Rises as flow falls | Confirm before doing it |
| Above roughly 8,000 | Axial, propeller | Flat, often with a dip | Rises steeply, maximum at shutoff | Do not; it overloads the motor |
The bands are approximate and different references draw them a few hundred apart, so treat a value near a boundary as a boundary case and get the pump's own power curve rather than arguing about which band it belongs to.
The power-direction column is the one that earns the calculation. A radial pump draws its least power at shutoff, which is why convention is to start one against a closed or nearly closed discharge and open it once the motor is up: minimum starting torque, minimum current. An axial pump draws its most power at shutoff, sometimes half again its rated power, so the same procedure is an overload every time. The pump did not change; the impeller shape did.
This also refines the sibling card on throttling, whose worked example is a radial machine. Throttling a radial pump reduces shaft power somewhat; throttling a mixed or axial pump can increase it, which makes throttling both expensive and hazardous to the motor on that end of the range.
The two corrections people drop
Stages. A five-stage vertical pump making 400 ft total is a machine of five 80 ft impellers. Divide first. Skipping this drops the computed value by a factor of about three and a third on a five-stage machine, and by the stage count to the three-quarter power in general, which will push a radial machine into a band it does not belong to, and will tell you the power curve runs the wrong way.
Double suction. Halve the flow before the square root. Missing it inflates the value by the square root of two, about 41 percent, which is enough to move a borderline machine one band. On a clearly radial machine it will not change the class, but it will make your number disagree with the manufacturer's stated value, and if you are using specific speed to compare two pumps for a duty, that disagreement is the whole comparison.
Filled shape card 1: a multistage vertical pump on a well
Inputs read off the curve sheet and the nameplate, with each correction on its own line.
- Shaft speed: 1,760 rpm
- Total head at BEP: 240 ft
- Stages: 3
- Head per stage correction: 240 / 3 = 80 ft
- Flow at BEP: 900 gpm
- Suction eyes: single, so flow per eye = 900 gpm, no correction
- Square root of flow: 30.0
- Head per stage to the 0.75 power: 80^0.75 = 26.75
- Specific speed: 1,760 x 30.0 / 26.75 = 1,970 (US customary units)
Read. Radial to the low end of Francis. Expect a reasonably steep head curve, a wide stable operating range, and shaft power that falls as flow falls. Starting this machine against a closed discharge valve is conventional and is easier on the motor than starting it open; open the valve promptly after the motor is at speed, and never leave a centrifugal shut in for more than a short interval, because the shaft keeps heating a trapped volume that can flash to steam when the valve is opened on a hot system.
Suction specific speed, same calculation with NPSH required in place of head. With NPSH required at BEP of 16 ft: 16^0.75 = 8.0, so 1,760 x 30.0 / 8.0 = 6,600. The industry commonly cites a ceiling somewhere in the 8,500 to 11,000 range, above which a pump's window of stable operation narrows sharply and part-load recirculation starts damaging the impeller; that ceiling is a design convention, not a regulation, and the pump manufacturer owns the real answer for the specific casing. At 6,600 this machine should tolerate part-load operation without a fight.
Filled shape card 2: a large stormwater pump
- Shaft speed: 590 rpm
- Total head at BEP: 14 ft
- Stages: 1, so head per stage = 14 ft
- Flow at BEP: 8,000 gpm
- Suction eyes: single, so flow per eye = 8,000 gpm
- Square root of flow: 89.44
- Head to the 0.75 power: 14^0.75 = 7.24
- Specific speed: 590 x 89.44 / 7.24 = 7,290 (US customary units)
Read. Mixed flow running into axial territory, which is what you would expect from a machine moving a lot of water a very short height. The head curve is flat, so small changes in system head produce large changes in flow. Shaft power rises as flow falls.
The operational consequences follow directly. Do not start this pump against a closed discharge valve; if the control scheme demands a closed-valve start, the motor has to be sized for shutoff power and the curve sheet has to say so, so get the power curve rather than assuming. Do not reduce its flow by throttling. Do not use a shut discharge as a quick way to stop it. And when it is commissioned, a wet well level that runs low enough to steepen the effective system head will move this pump's flow a long way for very little level change, which is a control problem the flat curve creates rather than a fault.
What would flip the read. If the same machine were re-rated at a lower speed for a smaller basin, recompute; specific speed is a function of the operating point, and a pump run well away from BEP does not behave like the shape card says. And if the curve sheet turned out to describe two stages rather than one, head per stage halves, the denominator drops, and the computed value rises further into axial territory, which strengthens rather than reverses the conclusion.
Where the number stops being useful
Specific speed describes centrifugal machines. It has no meaning for a positive displacement pump, whose flow is set by displacement and speed rather than by a curve, and a sibling card covers why that family cannot be throttled at all.
It also describes the shape, not the condition. A worn radial pump is still a radial pump: the calculation will not tell you the wear rings are gone. And it says nothing about the mechanical side, so it does not predict alignment, seal life or vibration.
Finally, it is computed at BEP and every conclusion above belongs to a machine operating reasonably near BEP. A pump running at a third of BEP flow is in internal recirculation regardless of its specific speed, and the shape card does not describe that condition.
How to verify you got this right
Compare your computed value against the manufacturer's published specific speed for the model if the curve sheet carries it, and if the two disagree by more than a few percent, the disagreement is almost always one of the two corrections rather than arithmetic. Check the stage count against the physical machine, not the model number, and count suction eyes by looking at the casing rather than trusting a name.
Then test the prediction with a clamp meter, since the power direction is the payload. On a radial machine, current should fall as the discharge valve is slowly closed; on a mixed or axial machine it should rise. Close in small increments with the suction valve fully open, watch the current on every increment, stop the moment it moves the wrong way against your prediction, and do not carry this test past a few seconds near shutoff on any hot system. If current rises on a pump you computed as radial, either the number is wrong or you are not looking at the pump you think you are.
References
- Hydraulic Institute standards for centrifugal pump nomenclature and specific speed, in the edition your specification or service contract references, which is also where the suction specific speed convention lives
- Pump manufacturer curve sheet for the specific casing, which owns the best efficiency point, the NPSH required, the shutoff power and the permissible starting arrangement
- See related: Reading Pump Curves Reference; Common Pump Types Reference; Why a Throttling Valve Is the Most Expensive Flow Control There Is; Why a Positive Displacement Pump Cannot Be Throttled