Why a Throttling Valve Is the Most Expensive Flow Control There Is

Why this matters

Every method of reducing flow through a pumped circuit costs something, and they do not cost the same thing. Throttling is the most expensive of the methods that genuinely reduce the flow the pump moves, and it is expensive for a structural reason rather than an incidental one: it is the only method that reduces flow by adding head, so the pump keeps making a large share of the power it made before and the valve turns the surplus into nothing. A diverting bypass is worse still on the same measure, which is exactly why it is not in the ranking - it does not reduce the flow the pump moves at all, and this card is largely about why the two are not comparable even though a customer will describe both as "turning it down."

A sibling card covers what a partly closed valve does to head, velocity and wear. This one compares the available methods on one number, at one duty, so you can tell a customer which one their money is going into.

Before you close anything on a discharge

Never throttle the discharge of a positive displacement pump. A gear, vane, piston, screw, progressing cavity or diaphragm metering pump does not slow down when you restrict it; it raises pressure until the relief lifts or something splits. Confirm which family of pump you are standing at before you touch a discharge valve, and if it is positive displacement, stop the pump instead of throttling it.

Do not use the discharge valve as a stop on a running centrifugal pump. Fully closed, the shaft keeps putting energy into a small trapped volume and the casing heats; on a system already hot, that water can flash when the valve is reopened. Shut the pump down first, then close the valve.

Where the valve you are handling is an isolation for service rather than a balancing adjustment, the mechanical and stored-energy requirements of 29 CFR 1910.147 apply: lock and tag the energy isolating device rather than switching it off, then bleed the trapped pressure at a gauge or vent and confirm the gauge reads zero before you break a joint. If the same job puts you inside the starter or the drive, that is electrical work under 29 CFR 1910.333(b)(2), with 29 CFR 1926.417 as the construction counterpart.

The measure, stated before any number is used

Comparing flow-control methods is easy to do dishonestly, because each method changes more than one quantity. So fix the comparison:

• At the same delivered flow. Not at the same pump flow, not at the same valve position. The duty is the constant.
• Shaft power at the pump, declared as such. Motor and drive losses sit outside these figures and get worse at part load, so the wire-side picture is slightly worse than every number below rather than better.
• On a friction-dominated system, which is stated because it is the condition the ranking was derived under. Static head changes the ordering, and it changes it enough to invert one entry. That fork is at the end of this card.

The output is a single ratio: shaft power as a fraction of design, divided by delivered flow as a fraction of design. At the design point that ratio is exactly 1.00 by construction. Anything below 1.00 means the method is moving each gallon more cheaply than it was moved at full duty; anything above 1.00 means each gallon now costs more than it used to.

What throttling actually does

Closing a valve steepens the system curve. The pump does not respond by working less; it responds by riding back up its own curve to a lower flow and a higher head. It is producing more head than the piping needs, and the valve absorbs the surplus and dissipates it as heat in the fluid.

The heat is not the story, and it is worth saying so plainly because people expect it to be. Head destroyed converts to fluid temperature rise at roughly one degree Fahrenheit per 778 ft of head for water, so 50 ft of destroyed head raises the fluid about 0.07 F, which no one will ever measure. The cost is the shaft power that never fell, not the temperature that never rose.

Worked comparison: one pump, one duty, three methods

Same pump used across this group. Published curve at full speed and full diameter: 104 ft at shutoff, 100 ft at 100 gpm, 96 ft at 150 gpm, 90 ft at 200 gpm, 82 ft at 250 gpm. Best efficiency near 200 gpm at about 78 percent, falling to roughly 70 percent at 140 gpm. The system is a closed friction-only loop designed at 200 gpm and 90 ft. A retrofit has reduced the load and the new duty is 140 gpm, which is 70 percent of design.

What the piping actually needs at 140 gpm. Friction head scales with the square of flow: 90 x (140/200)^2 = 44.1 ft.

Method 1: throttle a discharge valve

The pump rides its own curve to 140 gpm, where the curve gives 97.1 ft.

• Head produced: 97.1 ft
• Head the piping needs: 44.1 ft
• Head destroyed in the valve: 97.1 - 44.1 = 53.0 ft, which is 55 percent of everything the pump made
• Shaft power, with the efficiency correction the general section requires: (140 x 97.1 / 0.70) against design (200 x 90 / 0.78) = 0.84 of design power
• Power per unit of delivered flow: 0.84 / 0.70 = 1.20

Seventy percent of the flow at eighty-four percent of the power. Every gallon now costs 20 percent more to move than it did at full duty.

Method 2: trim the impeller

Trimming reduces the flow and the head together, which is the right shape of change. On a friction-only system the operating point tracks the same parabola, so reaching 140 gpm needs a diameter ratio near 0.70, a 30 percent cut.

That trim is not available. Manufacturers publish a minimum trim for each casing, commonly in the range of a 10 to 20 percent reduction from full diameter, below which the impeller no longer fills the volute properly and the published relationships stop predicting anything. The manufacturer's curve sheet owns that limit for the specific pump, not a rule of thumb. At a 15 percent trim this pump lands near 170 gpm, not 140, so trimming can carry part of this duty and cannot carry all of it. Note also that the trim relationships are approximations that get less reliable as the cut grows, unlike the speed relationships which are exact for the pump.

Where a trim does fit the duty, it is close to speed control in energy and beats throttling clearly. Its real cost is that it is permanent, and it is the wrong answer where the load will come back.

Method 3: reduce speed

On a friction-only loop, 70 percent flow is 70 percent speed and the cube law applies cleanly, because the reduced-speed operating point corresponds to the design point on the curve. The derivation is in the sibling card on variable speed; the result here is shaft power >= 0.343 of design, written as a bound because efficiency at the corresponding point is roughly but not exactly held.

• Power per unit of delivered flow: 0.343 / 0.70 = 0.49

Against throttling's 1.20, that is about 2.4 times less energy for the same gallons. That multiple is the honest headline of this card, and it belongs to this system and this duty rather than to pumping in general.

What is deliberately not on the list: the bypass

A three-way diverting valve at a coil, or a recirculation line back to the tank, gets described as flow control by everyone who installs one. Run it through the same measure.

The coil sees 140 gpm. The loop still sees 200 gpm, because the diverted 60 gpm goes around the coil rather than out of the system. The pump's operating point has barely moved, so shaft power stays near 1.00 of design.

• Power per unit of delivered flow: 1.00 / 0.70 = 1.43

That is worse than the throttle, and it is worse for a reason worth naming out loud: nothing was controlled. Flow was relocated. This is why a bypass is excluded from the ranking rather than placed at the bottom of it, and why the field evidence people offer for a bypass working ("the coil stopped overheating", "the noise went away") is evidence about the branch and says nothing about the pump. It is also why a system converted from three-way to two-way valves gets no energy benefit at all until something is done at the pump; the two-way valves make variable flow possible, and the pump has to be told about it.

The one case where a bypass is doing a real job rather than pretending to: protecting a pump against operating below its minimum continuous flow. There, dumping flow is the whole point and the energy is the price of not destroying the pump.

When throttling is still the right answer

Being the most expensive method per gallon does not make it the wrong choice.

• Balancing a branch against its neighbours. A balance valve is supposed to add resistance; that is its entire function. Comparing it to a drive is a category error, and the arithmetic above does not argue for removing balance valves.
• A small permanent trim on one branch of many. The energy at stake is a fraction of a fraction, and any capital method loses on effort alone.
• A temporary condition with a known end date, where a permanent trim would have to be undone.
• Where the duty change is under roughly 10 to 15 percent. The head destroyed is small because the pump has barely moved along its curve, and no other method is worth its disruption.

The condition that inverts the ranking: static head. On a system with substantial lift, reducing speed does not track the system curve, flow collapses far faster than speed and efficiency falls off a cliff, so speed control can cost more per delivered gallon than running full out. The sibling variable-speed card works that case with numbers. Before recommending a drive over a throttle, get the static fraction; it decides the answer, not the horsepower.

How to verify you got this right

Take the differential across the valve, not just across the pump. A balance valve or a triple-duty valve with a large pressure drop relative to the pump's total head is the physical evidence of the energy being destroyed, and it is a number a customer can see. A valve absorbing more than about half of the pump's differential is a system that is fighting itself, and it is worth pricing an alternative.

Then check the valve's own condition, because the throttle pays a second time. A valve held part open for years wire-draws its seat, and a valve taking a very large drop relative to its inlet pressure can cavitate across the trim, which erodes the plug and eventually gives you a valve that no longer closes when you need it to as an isolation. If the seat is cut, the fix is not a new valve in the same duty.

References

• Pump manufacturer curve sheet, which owns the efficiency values, the minimum permissible impeller trim and the minimum continuous flow for the specific casing
• Hydraulic Institute guidance on flow control methods for centrifugal pumps, in the edition your specification or service contract references
• 29 CFR 1910.147 for isolation and stored-energy release before breaking a joint at a valve, and 29 CFR 1910.333(b)(2) with 29 CFR 1926.417 as the construction counterpart for work in the starter or drive
• See related: Why Throttling a Valve Changes More Than Flow; What a Variable Speed Drive Changes About a Pumped System; Why a Positive Displacement Pump Cannot Be Throttled