Ready to Get Your VFD Solution?
Whether you need a single drive or a complete system, we have the VFD products and engineering support to ensure your project's success.
When to Use a VFD on a Pump (and When Not To)
Use a VFD on a pump when the system’s head is mostly friction and the flow actually varies. Do not use one when the head is mostly static lift, or when nothing is being throttled today. The number that decides is the static head vs friction head ratio, and it is worth working out before you price a drive.
The choice is not a detail. It is the difference between a project that pays back in eighteen months and one that never does.
Most pump VFD material starts with the pump affinity laws and a calculation showing power falling with the cube of speed. The laws are correct. Applying them to your energy estimate is where the trouble starts. They describe the pump, and say nothing about the system connected to it.
If the pump is already screened in and you need the drive sized, skip to VFD selection for pumps and fans. This article is about the decision that comes first.
Key Takeaways
- The pump affinity laws describe the pump curve, not the system curve. The Hydraulic Institute, Europump and the US Department of Energy jointly call using them for energy savings in a static-head system “a common mistake” that “can also lead to major errors.”
- A corrected power relationship replaces the cube law when static head is present. It reduces to the cube law only as static head approaches zero.
- In ORNL and DOE testing, one pump at the same reduced flow lost 1.2 percent efficiency in a no-static-head system and 4.6 percent in one with static head.
- A pump in a high-static system may not deliver flow until roughly 82 percent speed, and rangeability collapses to about 2:1 once static head passes roughly 30 percent of system head.
- Four cases rarely justify a drive: static-head-dominated systems, positive displacement pumps, pumps running constantly at rated flow, and systems with no throttling to remove.
Why the Cube Law Overstates Savings
What the Pump Affinity Laws Describe
For a fixed impeller diameter, the three relationships are:
Q2/Q1 = N2/N1
H2/H1 = (N2/N1)^2
P2/P1 = (N2/N1)^3Flow varies with speed, head with speed squared, and power with speed cubed. Run at 80 percent speed and the arithmetic suggests about half the power.
All three hold only while efficiency stays constant at corresponding points. ORNL and DOE put a number on it: above roughly 66.7 percent of full speed, it is normally acceptable to assume the new operating point has the same efficiency as the design point. Below that, degradation has to be modelled.
That figure is the most useful caveat here and it is missing from most published pump VFD guidance.
The Pump System Curve Does Not Start at Zero
A pump system curve plots the head needed to push a given flow through the installation. Friction head rises with flow, from pipe, fittings and valves. Static head is the lift that exists regardless of flow: the height a pump must raise water to, or the pressure it must hold against.
In a closed loop, static head is zero and the curve starts at the origin. Add lift, and the curve starts partway up the y-axis instead. The Hydraulic Institute, Europump and the US Department of Energy describe what follows in their joint guide on variable speed pumping:
“the system curve does not start from the origin but at some non-zero value on the y-axis corresponding to the static head. Hence, the system curve does not follow the curves of constant efficiency.”
And then the sentence that matters most when you are building a business case:
“A common mistake is to also use the Affinity Laws to calculate energy savings in systems with static head. Although this may be done as an approximation, it can also lead to major errors.”
The affinity laws move the pump down its own curve. With static head present, the operating point does not follow, because the system will not let the pump reduce its head below the lift.
The Corrected Power Relationship
ACEEE and ORNL publish a relationship that accounts for this:
P2/P1 = [1 - sqrt(H_min/H_N)] x (Q2/Q1) + sqrt(H_min/H_N)H_min is the static head and H_N the total head at design. As static head approaches zero, the expression reduces to the cube law. With meaningful static head it does not, and the gap widens as the ratio grows.
A Measured Comparison
ORNL and DOE measured this on a real pump rather than modelling it. At the same reduced flow of 900 gpm:
| System | Efficiency drop at the new operating point |
|---|---|
| No static head | 1.2 percent |
| With static head | 4.6 percent |
| Changed system curve | 5.4 percent |
The pump and the flow reduction are identical. The penalty is nearly four times larger with static head, and it lands exactly where the savings were supposed to come from.
Not sure how much of your head is static? Send us the pump curve and the system curve. We will work out the ratio and tell you what it means for the payback before you commit to a specification. Ask our engineers.
When to Use a VFD on a Pump: The Static Head Screen
Reading the Ratio Off a System Curve
You need two numbers: static head, and total head at design flow. Both are usually on the pump schedule.
static head share = H_static / H_total at design flowA pump lifting from a wet well to a discharge 30 metres above, with total head of 40 metres at design, has a 75 percent static share. That system is a poor VFD candidate regardless of how attractive the cube law looked.
What the Numbers Mean
| Static head as a share of total head | Assessment |
|---|---|
| Under 20 percent | Strong candidate. Savings approach the affinity-law estimate |
| 20 to 50 percent | Credible candidate. Model the actual operating point |
| Over 50 percent | Usually not worth a drive. Check alternatives first |
The 50 percent line is a screening rule from pump training material, not a formal standard. No Hydraulic Institute publication or manufacturer standard states a numeric cutoff. The documented DOE position is qualitative: it states that drives “must be evaluated carefully for use in systems that have high static head.”
A related figure to be sceptical of. A friction-head threshold of 30 to 40 percent is commonly cited as the level above which a VFD becomes worthwhile. It appears in no standard I could find, and the documented guidance runs the other way. It is high static head that disqualifies a system, not low friction head.
Why Rangeability Collapses
Static head caps how far you can turn the pump down. A pump in a high-static system may not deliver any flow until roughly 82 percent speed. Below that it cannot generate enough head to overcome the lift, and the operating point falls off the curve.
ISA states the principle plainly: “the magnitude of the static head determines how low in speed a VSD can go.” ISA also flags the failure mode. If pump head drops below static head, reverse flow through the pump can occur.
So a high-static system may give you 2:1 flow turndown rather than the 5:1 or better a friction-dominated system achieves. Less turndown means less time at reduced speed, which means less energy saved.
Selin operates a municipal water utility in the Aegean region of Turkey and inherited a VFD proposal for a booster station serving a hillside district. The quote promised a 40 percent saving from a 60 percent speed reduction. Static lift to the district was 62 metres against a total head of 78. When her engineer plotted the system curve, the pump could not have reached 60 percent speed and produced flow at all. Realistic turndown was about 84 percent, the saving was closer to 11 percent, and payback moved from under two years to over nine. The proposal had not been dishonest. It had applied a closed-loop formula to an open system.
Four Pumps That Should Not Get a VFD
Static-Head-Dominated Systems
Covered above, but worth stating as a decision. Where static head is most of the total, reducing speed buys very little before the pump stops producing flow.
DOE also notes these pumps run close to shut-off head conditions, producing “greater shaft deflection, high vibration levels, and high bearing loads.” The drive is not merely ineffective there. It is a route to mechanical damage.
Positive Displacement Pumps
The cube law does not apply. ABB’s guidance shows PD pumps have linear power against speed, and the Hydraulic Institute guide explains why:
“For a PD pump, the flow is proportional to speed, but the pressure can be independent of speed. Consequently, power and energy savings do not fall so quickly when speed is reduced.”
A PD pump still benefits from a VFD for process control, soft starting and pressure regulation. It rarely benefits much on energy.
Pumps Running Constantly at Rated Flow
ABB is direct: variable speed control “is not recommended for pumping systems almost constantly operated at the nominal flow rate.” The drive is not free. At full flow its own losses make the system roughly 3 percent less efficient than running the pump directly, and a pump that never turns down pays that penalty permanently.
Systems Where Nothing Is Being Throttled Today
A VFD saves energy by replacing a loss you already pay for, usually a part-closed valve or a bypass. If the pump runs at its required flow with the valve wide open, there is no loss to remove.
The Hydraulic Institute sets out the order of remedies, and the drive is last: trim the impeller, then replace the pump, then install a variable speed drive. The first two reduce the loss. The third only recovers it.
If your pump is simply oversized, an impeller trim or a correctly sized replacement may deliver most of the saving at a fraction of the cost. Send us the pump curve and we will tell you which remedy the numbers actually support.
What Else Has to Be True Before You Specify
Minimum Continuous Flow and Recirculation
Every centrifugal pump has a minimum flow below which the manufacturer will not warrant operation. DOE defines it as “the lowest flow rate at which the pump can operate without risking damage from suction or discharge recirculation,” and recirculation causes what DOE calls “cavitation-like damage.” A Goulds manual places suction recirculation onset below 50 to 75 percent of BEP flow.
Minimum flow sets a floor on turndown, and it often binds before the static-head limit does.
Minimum Speed, and Why There Is No Universal Number
You will see “30 Hz” quoted as the minimum speed for a pump VFD, and it is real in real manuals. Sulzer’s VMS vertical multistage IOM lists 30 Hz as the minimum frequency. Grundfos documents 30 percent of nominal in some pump families and 50 percent in others. Franklin Electric specifies 30 Hz for submersible motors on bearing-lubrication grounds.
There is no universal number. Minimum speed is set by the speed needed to produce flow against static head, by motor cooling at low speed, by bearing lubrication, and by minimum continuous stable flow. Take it from the pump and motor documentation.
NPSH at Reduced Speed
Reducing speed generally helps suction performance. The Hydraulic Institute states it directly: “Increasing pump speed will negatively affect pump suction performance. Conversely, reducing speed will have a positive effect.”
NPSH required scales roughly with speed squared, so slower means more margin. The exception: past the best efficiency point the NPSHr curve rises so steeply that a lower-speed curve can require more NPSHr than a higher-speed one at the same flow.
Where the Operating Point Sits Against BEP
ANSI/HI 9.6.3 defines a preferred operating region around the best efficiency point: 70 to 120 percent of BEP flow for specific speeds at or below 87, 80 to 120 percent at higher specific speeds, and 80 to 115 percent for high-energy API 610 pumps.
A VFD rule that follows: for a system with some static head, select the pump so maximum flow sits slightly to the right of BEP. The exception is a constant-pressure regulated system, where the pump should sit left of BEP at maximum pressure. DOE’s screening tip suggests prioritizing pumps already running 30 percent or more from BEP.
Selecting the Drive Once the Screen Passes
Current, Not Horsepower
Size on the motor’s full-load current, not nameplate kW, and match the drive’s continuous output current to it at the duty class the load needs. Centrifugal pumps are variable torque, which usually means a normal-duty rating. The method is set out in how to size a VFD for a motor.
Control Mode: V/f Is Normally Enough
You do not need vector control for an ordinary centrifugal pump. ISA states that open-loop volts-per-hertz control “has the simplest algorithm but is susceptible to varying degrees of slip. Most of the drives provided for pump control use this strategy.”
ABB agrees, placing scalar control with “some pumps and fans” at plus or minus 2 to 3 percent speed accuracy. Vector earns its place where you need starting torque above 100 percent rated, regulation tighter than 0.2 percent, torque limiting, or smooth running at very low speed.
Where the Sizing and Configuration Methods Live
This article stops at the screen. VFD selection by load type covers load classification and drive type, and configuring a VFD for pumps and fans covers parameter setup.
Multi-Pump Stations
One VFD per pump is the standard arrangement. A shared drive becomes a single point of failure for the station and constrains how the pumps run together.
If you do run several motors from one drive: all motors start and stop together at the same speed, the drive is sized on summed full-load current, V/f is required because sensorless vector is impossible across multiple motors, and each motor needs its own overload protection because the drive’s thermal model only sees combined current.
Xylem’s controller documentation gives the vocabulary worth specifying: Stage Speed sets the point a lag pump starts, Destage Speed the point one stops, and alternation runs on a timed, daily, weekly or monthly cycle, sequenced by pump number or by run hours.
One counterintuitive point from DOE: parallel pumps suit high-static-head systems better than speed control does, because staging keeps each pump nearer its BEP. Where static head rules out a drive, adding a pump may be the better answer.
Ramps, Surge and Water Hammer
A VFD usually reduces surge risk. The Hydraulic Institute notes “there is less likelihood of flow or pressure surges when the control device provides rates of change, which are virtually infinitely variable.”
But it can make surge worse if the ramp is set badly. Rockwell documents the mechanism: excessive acceleration torque brings the pump to speed too quickly and produces a flow surge at the end of the start cycle. A stop ramp set too short is effectively a rapid valve closure. The remedy is torque-controlled acceleration and deceleration rather than voltage ramping alone, plus the pump-specific features built for it. Danfoss exposes a Check Valve Ramp that slows ramp-down as the check valve nears its seat, and a Pipe Fill Mode for filling empty lines without bursting them.
There is no universal ramp time. Product defaults such as 20 seconds to start and 10 to stop are starting points, not standards. The right value depends on pipe length, fluid inertia and the pressure rise a given rate of flow change produces.
Building the Payback Case Honestly
Three inputs decide whether the arithmetic survives contact with the plant.
The real hours at reduced speed. A pump that will not produce flow below 82 percent speed has a narrow turndown band. Model the hours actually spent there, not a notional speed profile.
The component losses. DOE notes that “anticipated energy savings are not realized in some applications because some of the losses associated with VFD installation were not taken into consideration.” Drives are about 95 to 97 percent efficient, and motor efficiency generally starts falling below 75 percent load.
Wire-to-water efficiency is measured as a baseline. This is standard DOE vocabulary for overall efficiency from electrical input to hydraulic output:
eta_wtw = eta_pump x eta_motor x eta_vfdTypical values are a pump at 70 to 85 percent, a premium motor at 93 to 96 percent, and a drive at 96 to 98 percent, giving roughly 60 to 75 percent overall. DOE’s pumping tip sheet recommends testing it and keeping records to spot trends, noting wear ring and rotor erosion can reduce it by 10 percent or more.
Measure it before the drive goes in. Without a baseline you cannot demonstrate the savings afterwards.
Bhavesh manages utilities at a chemical plant in Gujarat and installed a VFD on a cooling water pump against a calculated 28 percent saving. The first year delivered 9 percent. Nothing was wrong with the drive. The calculation assumed the pump would spend most of its time at 70 percent speed, and process constraints kept it above 88 percent for all but a few weeks a year. A year of logged speed data beforehand would have predicted it, and the project would still have gone ahead, at a stated payback of five years rather than eighteen months.
Frequently Asked Questions
Is a VFD worth it for a pump?
When the system is mostly friction head, the flow actually varies, and something is being throttled today. The stronger the static head share, the weaker the case.
What is the static head limit for a pump VFD?
A widely used screening rule treats static head above roughly 50 percent of total head as a poor candidate. That figure comes from pump training material rather than a formal standard, so use it as a filter and verify against the actual system curve.
How much energy does a VFD save on a pump?
It depends almost entirely on the system, not the pump. In a friction-dominated system the savings approach the affinity-law estimate. In a high-static system it can fall to single digits, bounded by how far the pump can turn down before it stops producing flow.
What is the minimum speed for a pump VFD?
There is no universal figure. Sulzer documents 30 Hz for its VMS range, Grundfos documents 30 percent of nominal for some families and 50 percent for others, and Franklin Electric specifies 30 Hz for submersible motors. The real limit is the speed at which the pump still produces flow against static head while staying above minimum continuous flow.
Why does the affinity law overestimate my savings?
Because it describes the pump, not the system. With static head present the operating point does not follow the pump down its curve, so the pump loses efficiency faster than the cube law assumes. The Hydraulic Institute and DOE both identify this as a common error.
Does running a pump slower reduce cavitation risk?
Usually yes. NPSH required falls roughly with speed squared, so slower speed increases suction margin near the best efficiency point. Past BEP the NPSHr curve rises steeply enough that a lower-speed curve can need more NPSH than a higher-speed one at the same flow.
Conclusion
The question is not whether a VFD can control a pump. It can. The question is whether the system rewards it.
Work out the static head share of total head first. Under 20 percent is a strong candidate. Over 50 percent usually is not. In between, model the actual operating point rather than reaching for the cube law, and check how far the pump turns down before it stops producing flow.
Then clear the four operating limits: minimum continuous flow, minimum speed, NPSH margin, and position against the best efficiency point. A drive that satisfies the energy case but violates one of those is not a project, it is a warranty claim.
Only then size the drive on current rather than horsepower.
