Throttling reduces flow by adding resistance: the pump rides up its curve to a higher head, and the surplus head is burned off as heat across the valve. Flow drops a lot; power drops only a little.
Variable speed (a VFD) reduces flow by lowering the whole pump curve. The operating point slides down the real system curve — no wasted head — and by the affinity laws power falls with the cube of speed.
The saving is largest on friction-dominated systems and smallest on high-static-head systems, where the pump can only slow so far before it can no longer lift the liquid at all.
1 · Two ways to turn a pump down
Say a pump is delivering its full design flow and you need less — a process turndown, a cooler night, a tank nearly full. You have two levers, and they move the operating point in completely different directions.
Throttle the discharge valve. Closing the valve adds resistance, which steepens the system curve. The pump keeps spinning at full speed, so it stays on its full-speed curve — and the new operating point slides up and to the left: less flow, but higher head. The pump is now making more head than the pipework needs; the excess is dropped across the valve and turns into heat and noise. You met the flow target by deliberately wasting energy.
Slow the pump with a VFD. A variable frequency drive lowers the supply frequency, so the pump turns slower and its whole curve drops. The system curve doesn't move (the valve stays wide open), so the operating point slides down the real system curve: less flow at exactly the head the system needs, and not a metre more. Nothing is thrown away.
2 · Why the VFD wins — the cube law
We met the affinity laws in Part 1; this is where they pay off. For a fixed pump, changing speed N scales the curve by simple ratios:
Throttling gets none of this. Because the pump stays at full speed, it keeps drawing close to full power even as the flow falls — the hydraulic power ρgQH barely drops, because H rises as fast as Q falls, and then most of what's left is wasted at the valve anyway. The gap between the two curves is the energy you are paying for and throwing away.
This is the same ρgHQ work argument from Part 1, seen from the cost side. The throttle valve does no useful work — it only destroys head. A VFD removes the need to make that head at all.
3 · See the energy gap
Set a turndown — how far below full flow you need to run — and the system's static-head fraction. The left chart shows where each method lands on the head-flow plane; the red shaded band is the head thrown away at the throttle valve. The right chart is the one your finance team cares about: power drawn versus flow for each method, with the gap between them being your saving.
Interactive — Throttle vs VFD energy & cost
Live modelWhere each method operates
Power vs flow — the bill
H = s²·50 − 0.0009·Q²; power is electrical (pump efficiency off-BEP, 92% motor, 97% drive). Throttle stays at full speed and dumps surplus head across the valve; VFD slows to meet the system curve. Indicative for teaching — real savings need the actual duty profile and certified curves.4 · The catch: static head
Drag the static-head fraction up in the model and watch the VFD's advantage shrink. This is the most important caveat in the whole subject, and it is where a lot of disappointing VFD retrofits come from.
A pump can only slow down so far. Its head falls with the square of speed, so at some reduced speed the pump can no longer produce even the static head — the pure lift — and flow collapses to zero. On a system that is mostly static lift (pumping up a tall column with little pipe friction), there is very little speed range to play with, so the VFD saving is modest. On a system that is mostly friction (a closed circulation loop, a long pipeline), the static head is small, the pump can slow right down, and the cube law delivers dramatic savings.
| System type | Static head | VFD energy saving | Examples |
|---|---|---|---|
| Friction-dominated | Low | Large (often 40–70%) | Closed cooling loops, circulation, long transfer lines |
| Mixed | Medium | Moderate | Most general transfer duties |
| Static-dominated | High | Small | Lifting to a high tank with short pipework |
5 · Beyond energy
Energy is the headline, but a VFD brings more — and a few costs to weigh:
- Soft starting — ramping frequency means full torque at low current, as covered in Motor starting methods: no inrush, no water hammer.
- Process control — speed becomes a smooth, continuous control variable for pressure, flow or level, far better than a control valve.
- Reduced wear — running slower at part load lowers vibration and extends seal and bearing life.
- Costs & caveats — drives cost money, generate harmonics that may need filtering, can struggle below ~25–30% speed, and need a minimum-flow check so the pump never drops below its safe continuous flow.
The honest design rule: if the duty varies and the system has meaningful friction, a VFD almost always pays — in energy, reliability and control. If the duty is fixed and the system is static-dominated, size the pump correctly (see Part 3) and a VFD may be hard to justify on energy alone.
Key takeaways
- Throttling reduces flow by wasting head across a valve; the pump stays at full speed and full-ish power.
- A VFD reduces flow by lowering the curve, sliding down the system with no wasted head — and
P ∝ N³. - Savings are largest on friction systems, smallest on static-head systems, where minimum speed limits the turndown.
- A VFD also soft-starts, controls and protects — but costs money and adds harmonics and a minimum-flow concern.
- Don't oversize and throttle for life — size correctly, and use speed for the variation.