The 40 to 60 percent number is back. A story about drives as hidden grid capacity ran in IEEE Spectrum in the middle of October, and it says a drive can cut a motor’s energy use by around 40 to 60 percent, with nothing in the piece about what the motor is turning. That figure is real on the right load, and it is also the figure that ends up on a project justification for a conveyor, where it was never true.
Nobody in maintenance needs to be talked out of drives. The trouble is that a savings estimate built on the wrong load type gets checked a year later against the power bill, comes up short, and the next drive project at that plant gets a harder look than it deserves. So the number is worth taking apart before it goes on a form.
Where the big number comes from
Centrifugal pumps, fans, and centrifugal compressors are variable torque loads, and they follow the affinity laws. Flow goes with speed, pressure goes with the square of speed, and the power they draw goes with the cube of speed. That cube is where the money is. At 80 percent speed a fan needs about 51 percent of its full speed power, and at 50 percent speed it needs about 12.5 percent.
The same reporting chain that carried the 40 to 60 percent figure also printed that slowing a fan by 25 percent cuts its power by roughly 58 percent, which is just the cube worked out (0.75 x 0.75 x 0.75 is about 0.42). That is a correct number for one fan at one operating point. A few paragraphs later the same article says the estimated savings for fans and refrigeration compressors across the whole country came out closer to 9 percent. Both numbers are defensible. They are 6 times apart because one is the physics at a single speed and the other is what happens across a fleet of real machines that spend most of their hours somewhere other than the best case.
Pumps carry one more catch. A pump that lifts water up a hill, or pushes against a tank held at pressure, has static head that does not go away when the pump slows down, so a good part of what it does at full speed it still has to do at reduced speed. The cube law describes friction head. On a system with a lot of static head the savings curve flattens out, and sometimes the pump cannot slow down very far at all before it stops moving water.
Constant torque loads do not get the cube
Conveyors, mixers, extruders, hoists, positive displacement pumps, and reciprocating compressors need about the same torque at low speed as they do at full speed. Power is torque times speed, so on these loads power falls in a straight line with speed and there is no cubic windfall to collect. Slowing a conveyor by 20 percent takes about 20 percent off its power, and only if the process actually needed it to run slower.
The positive displacement pump is the one that catches people, because it is a pump and everybody has learned that pumps are the drive candidates. A gear pump, a lobe pump, or a progressive cavity pump moves a fixed volume per turn and builds whatever pressure it takes to push that volume through the system. It is a constant torque machine wearing a pump nameplate, and putting the 40 to 60 percent figure on one is a mistake that shows up on the power bill.
Drives earn their keep on constant torque loads for other reasons: soft starting, controlled acceleration, process speed control, less wear on belts and gearboxes, and a much lower inrush than starting across the line. Those are real and they are worth money. They just need to be written down as what they are instead of as energy savings.
Even on pumps, how it runs decides the answer
The best public field data on this comes from NEMA working with the Northwest Energy Efficiency Alliance and Cadeo Group, who took measurements from 132 clean water pumps in the Pacific Northwest and modeled each one two ways: speed control against throttling a valve at constant speed. They split the pumps by how they actually ran. A pump that spent 90 percent or more of its time at one operating point was called constant load, and anything that spent real time at two or more points was variable load.
Variable load pumps came out around 43 percent savings. Constant load pumps came out around 23 percent, mostly because a drive lets an oversized pump be slowed down to the job it really has instead of being choked back with a valve. The municipal water pump in the study showed zero. It was set up as a constant speed pump running at its duty point with a separate variable speed trim pump handling the swings, so there was nothing left for a drive on the big pump to recover.
Two things worth noticing about that study. First, every machine in it was a pump, so it says nothing directly about conveyors or mixers. Second, the same kind of machine gave 43, 23, or 0 percent depending on how the system around it was run. The savings live in the operating profile, and the operating profile is something you can go look at.
What to check before the number goes on the form
Before you estimate savings on a drive project, get five things on paper.
- The load type. Centrifugal pump, fan, or centrifugal compressor is variable torque. Conveyor, mixer, extruder, hoist, positive displacement pump, or reciprocating compressor is constant torque.
- How it is controlled today. Throttling valve, discharge damper, inlet vanes, bypass line, or on and off cycling each leave a different amount on the table.
- Hours at each operating point. A valve position trend, a damper position, a flow meter history, or even a week of current readings on the motor will tell you more than a nameplate.
- Static head on pump systems, because that sets how far the pump can slow down and still do its job.
- Whether anything else is already doing the speed control, like the trim pump on that municipal system.
With those in hand you can run the affinity laws against the real hours instead of against the best case, and the estimate you hand in will still look right after a year of power bills. Without them, the honest thing to write is that the savings depend on how the load runs and that it has not been measured yet.
If the load spends most of its time at low speed, the motor’s cooling at reduced speed is the next question, which is covered in How Much Cooling Does a Motor Lose at Reduced Speed?. Sizing the drive for the load you actually have is in the VFD Sizing Guide, and the longer treatment of load types and torque curves is in Before the First Fault. If your team needs to be able to make this call on their own, that is what the training is for.
Frequently Asked Questions
How much energy does a VFD save?
It depends on the load. On a centrifugal pump or fan that spends a lot of time at reduced flow, savings of 40 percent or more are possible because power falls with the cube of speed. On a pump that sits at one operating point most of the time, field data put it closer to 23 percent, and on a constant torque load like a conveyor the saving is roughly proportional to how much you slow it down.
Does a VFD save energy on a conveyor?
Only in proportion to how much slower it runs, because a conveyor needs about the same torque at every speed and its power falls in a straight line with speed. Drives on conveyors are usually justified by soft starting, speed control, and reduced mechanical wear rather than by energy.
Why does a positive displacement pump not save as much as a centrifugal pump?
A positive displacement pump moves a fixed volume every revolution and builds whatever pressure it takes to push that volume. That makes it a constant torque load, so it does not follow the cube law that gives centrifugal pumps their big savings.
Can a VFD save energy on a pump that runs at constant flow?
Often yes, if the pump is oversized and is being held back with a throttling valve, because the drive lets you slow it down to the flow it really needs. In field data on pumps that held one operating point most of the time, the modeled saving was around 23 percent. If the pump is already running right at its duty point with no throttling, there may be nothing to save.
What information do I need to estimate VFD energy savings?
The load type, the current control method, hours spent at each operating point, and on pump systems the static head. A trend of valve or damper position, or a week of motor current readings, is worth more than the nameplate for this.
Online start-up courses, filmed on real drives
One drive per course, filmed uncut with a clock on screen from first power to a saved backup. The install course is free, and the ABB ACS380 and ACS580 start-up course is open now.