BoreJet

Eductor Selection: Sizing, Pressure, Materials, and the Pump Trade-Off

RCRay Chan·August 25, 2026
Eductor Selection: Sizing, Pressure, Materials, and the Pump Trade-Off
Table of Contents

An eductor looks like a pipe fitting, is priced like one, and quietly does the work of equipment that costs ten times as much. It has no motor, no impeller, no shaft seal and no moving parts in the tank: a small pumped stream, the motive flow, drags several times its own volume of surrounding liquid along with it, which is why one modest pump can circulate a whole tank. But that power is conditional. An eductor only earns its keep when the duty fits its operating window: enough motive pressure, low discharge back-pressure, a fluid thin enough to be entrained, and an installation that keeps the body submerged and the discharge aimed where it matters.

This guide treats eductor selection as a decision process, not a catalog lookup: when an eductor beats a pump, the five parameters that fix every selection, a duty-based selection table, a five-year cost comparison against mechanical pumps, and the mistakes that show up on commissioning day. The numbers come from published engineering bulletins and vendor selection guides, and are deliberately presented as ranges. An eductor is specified against your worst case, not the vendor’s best case.

Eductor vs Pump: The Decision That Comes First

Before any sizing math, decide whether an eductor belongs in the system at all. The wrong device, perfectly sized, is still the wrong device. An eductor is a momentum multiplier, not a pump: it converts the pressure energy of a small stream into bulk movement of a large one, and only where the two streams live in the same body of liquid, or close to it.

Choose an eductor when the job is recirculation, blending or suspension inside a vessel, dilution of one liquid into another, lifting liquid from a sump, or modest transfer between nearby points. Choose a pump when the job is moving liquid over distance, against significant head, or to a precise metered destination. An eductor cannot push liquid up a long vertical riser the way a centrifugal pump does, nor dose a chemical to a fixed rate. It entrains whatever is available around its suction ports.

Criterion Eductor loop Mechanical pump loop
What it moves Motive flow plus 3–5× entrained flow Exactly what the pump delivers
Best duty In-tank circulation, blending, sump lift, dilution Transfer, high head, metering, long piping runs
Moving parts None at the point of use Impeller, shaft, seals
Pressure it needs 0.7–4.8 bar at the body, typically Whatever the system curve demands
Volume multiplied Yes: entrainment ratio 3:1 to 5:1 No: one liter per liter
Energy cost Small pump at moderate pressure Full flow at full head
Failure mode Silent: suction collapses, mixing stops Visible: cavitates, leaks or trips

The boundary cases are where the decision actually matters. A 20-meter recirculation line back to a tank: the pump loop needs a pump sized for the whole recirculation flow; the eductor loop needs one sized for the motive stream only, roughly a quarter to a fifth of the circulation flow. A sump emptied against 4 meters of lift: a jet-pump arrangement with a small external pump does it without submerging a pump in the pit: a real advantage where solids or aggressive liquid live. A 200-meter cross-plant transfer line is a pump’s job. The general rule: if the liquid ends up back in the same vessel or an adjacent one, an eductor loop is worth evaluating; if it is going somewhere else, buy the pump.

The Five Parameters That Fix Every Eductor Selection

Every eductor specification collapses into five numbers and one drawing: the liquid flow the system must circulate, the motive pressure available at the body, the flow amplification ratio the body can deliver at that pressure, the materials the fluid demands, and the installation geometry: submergence, aiming and spacing. Change any one and the selection changes.

  • Liquid flow: tank volumes per hour, converted to circulation, divided by the amplification ratio for motive flow.
  • Pressure ratio: the differential between motive supply and discharge at the body, not the nameplate.
  • Flow amplification ratio: induced flow divided by motive flow, typically 3:1–5:1 in liquid service.
  • Materials: driven by corrosion, erosion, temperature and rating, never by habit.
  • Installation: submergence below operating low level, discharge aimed along the floor or long axis, mutually reinforcing spacing.

The sections that follow take them in that order, flow, pressure, ratio, materials, installation.

Step One: Define the Liquid Flow and Turnover You Need

Eductor selection starts from the tank, not the catalog. Write down the vessel volume, decide how many complete tank volumes must circulate per hour, multiply. That number, the circulation requirement, is the master input to everything else.

For most blending, equalization and suspension duties, three to six tank volumes per hour is the practical band; pH equalization with slow reagents can run at one or two; suspension of settling solids usually needs the aggressive end, four to six, with the discharge aimed to keep the settled bed moving. A worked example fixes the method: a 20 m³ process tank at four turnovers per hour needs 80 m³/h of circulation: 80,000 liters per hour moving through the tank.

That circulation does not all have to come from the pump. The pump only supplies the motive stream, and the eductor multiplies it. At a flow amplification ratio of 4:1, 80 m³/h of circulation needs 20 m³/h of motive flow; at 5:1 it needs only 16 m³/h. The sizing rule published by tank mixing eductor manufacturers is the same arithmetic in one step: divide the required circulation by the amplification ratio, commonly taken as 5 for a first estimate, to get the needed inlet flow. Twenty m³/h of motive flow at 2–3 bar is a modest pump, a fraction of the size of the 80 m³/h pump a straight recirculation loop would demand.

Two refinements before the numbers go into a purchase order. First, size for the worst-case turnover, not the average. A tank that must be uniform within five minutes of a chemical dump needs the turnover that achieves that, and a loop sized for gentler duty will not catch up later. Second, decide between several smaller bodies and one large one: multiple eductors on a common header divide the motive flow, spread the plumes, and survive blockage of one body. The usual starting layout: one eductor per 2–3 m of tank length, all aimed in the same rotational direction.

Step Two: Stay Inside the Motive Pressure Window

The pressure parameter is where most selection errors live, because it is the one number measured at the wrong place: the differential at the eductor itself, between motive inlet and discharge, not the pump’s nameplate pressure.

Published tank mixing data sets the practical window: Spraying Systems rates its standard tank mixing eductors for roughly 10–50 psi (0.7–3.4 bar) of motive pressure, the classic Penberthy-style jet pump bulletin gives 10–70 psi (0.7–4.8 bar), and some heavy-duty lines are rated to 10–150 psig. Within the window, performance is pressure-driven: higher motive pressure produces a faster jet, a deeper suction and a higher flow amplification ratio: up to the body’s design point, beyond which gains flatten. Below the window’s lower edge, the device fails the quiet way: it passes motive flow but stops pulling suction. Mazzei, the venturi injection specialist, states it plainly: each injector has minimum flow and minimum pressure differential to initiate suction; if either is not met, water flows through but no suction is created.

The selection consequence is a test that avoids most field failures: read the pump curve at the eductor, after pipe friction, fittings and elevation loss in the motive line, and confirm the result sits inside the body’s window with margin. A pump delivering 5 bar at its flange can present 3 bar at an eductor 20 meters of pipe away, and a body rated for 10–50 psi sitting at the low end of its range will entrain a fraction of its rated flow. If the available pressure is marginal, the options in order of cost: a shorter or larger motive line, a lower-minimum-pressure body, or a booster pump.

The second pressure number is the discharge back-pressure. Liquid head above the discharge point, a tall tank, or a long discharge line all raise back-pressure and all suppress the amplification ratio. The NCI eductor selection guide gives roughly one foot of effective plume per psi of differential pressure. The practical reading: a body near the top of a tank with half the tank’s head on its discharge is a different device from the same body near the floor. Size against the discharge pressure you will see at the worst tank level.

Step Three: Lock the Flow Amplification Ratio to the Worst Case

The flow amplification ratio, also called the entrainment or induction ratio, is induced flow divided by motive flow, and it justifies the whole installation. One unit of pumped liquid dragging three to five units of tank liquid is the entire business case; without it, an eductor is just an inefficient pipe.

Published values are consistent across independent sources. Spraying Systems states that eductors entrain up to five times the pumped flow and that its standard models circulate four to six times the inlet flow; its sizing procedure divides the required circulation by 5. The Penberthy O-1660 data states that within 10–70 psi, four gallons of tank contents are mixed for every gallon of motive flow, so discharge is five times the inlet. The NCI selection guide lists a 4:1 suction pick-up ratio across its liquid-eductor range. The engineering consensus for liquid-liquid service is 3:1 to 5:1, and a responsible selection is made at the bottom of that band, not the top.

What drags the ratio down matters as much as the number itself. Back-pressure suppresses entrainment as tank level rises; viscosity suppresses it too, which is why practical selection guides bound liquid service at roughly 1 to 2,000 cP, above that, induction collapses and a mechanical mixer is usually the honest answer. Submergence matters: if the suction ports draw air instead of liquid, the ratio falls toward zero. And the ratio is not constant across a catalog line. A smaller nozzle at the same pressure gives a faster jet and a deeper vacuum but passes less motive flow, while a larger throat accepts more induced flow but weakens the vacuum. The nozzle-to-throat area ratio for liquid-liquid service is typically 2 to 4.

The selection discipline is to size the body for the lowest ratio you will see, worst-case tank level, thickest liquid, longest discharge run, and treat the better days as margin. A system specified at 3:1 that actually runs at 4.5:1 costs nothing extra and mixes faster; a system specified at 5:1 that actually runs at 3:1 never meets its turnover spec, and the only fix is buying more bodies.

Step Four: Match Materials to the Fluid, Not to Habit

Because an eductor has no seals and no rotating parts, material selection is simpler than a pump’s, nothing to lubricate, pack or replace, but it is not free. The fluid decides the material, and the two failure mechanisms are corrosion and erosion.

Clean water service is satisfied by brass or bronze bodies: the economy choice for plating rinse tanks and general recirculation, rated in the 500 psig class at ambient temperature in published selection data. Carbon steel handles neutral process liquids and higher pressures (about 740 psig at 100 °F in the same data). 316 stainless steel covers food, beverage, pharmaceutical and aggressive chemical service at about 720 psig at 100 °F: the default when the liquid is anything the plant would not drink. Plastics, PVC, polypropylene and, for the most aggressive chemistry, PVDF or PTFE-lined construction, handle acids, caustics and water-treatment chemicals at lower pressure ratings, which is acceptable because those services rarely need high pressure anyway.

Erosion changes the picture even when corrosion does not. In abrasive service, slurries, recirculated water carrying fines, the nozzle bore wears first, because that is where the liquid reaches its highest velocity. A worn nozzle enlarges the bore, drops the jet velocity, and quietly halves the amplification ratio before anyone notices. The defense is a replaceable nozzle insert in a harder material, plus a spare on the shelf: the upgrade that pays back fastest.

Temperature and pressure rating travel together. Plastics lose strength as temperature rises, and the ratings quoted above are ambient-temperature numbers, if the liquid is hot, check the derating curve before trusting the nominal rating. Stainless and carbon steel hold their ratings across a much wider temperature band, which is why hot process duty usually ends up in metal regardless of what the chemistry alone would allow.

Step Five: Design the Installation, Not Just the Part

An eductor correctly specified and wrongly installed is a failed project. Installation is the most under-designed parameter. Three facts govern: the body must run submerged, the discharge aimed where mixing is needed, and the manifold feeding every body at uniform pressure.

Submergence is non-negotiable. The suction ports need liquid above them at the worst-case operating level. Level swings are the classic cause of a system that mixes on Monday and gurgles on Friday. Mount the body below the operating low level; if the tank has a deep settled solids layer, keep the suction ports above the settled bed while aiming the discharge to keep that bed in motion.

Aiming does the actual mixing. The greatest agitation is in the discharge plume, so aim the discharge end at the most remote part of the tank: parallel to the floor or slightly downward, across the long axis. A discharge aimed upward churns the surface and leaves the bottom stratified. In cylindrical, square or rectangular tanks, corners are where flow stalls; the aiming pattern keeps them in motion.

The manifold matters as much as the bodies. The supply line and header must be sized so each body sees the same pressure: a starved eductor at the end of an undersized header silently under-performs while bodies near the pump run at full ratio. Purge and vent the header before start-up: air locks kill individual bodies while the rest appear to work. Log the commissioning baseline, turnover time, pressure at each body, plume pattern, because six months later, when mixing has quietly degraded, it tells you whether the nozzle wore, the header starved, or a repair re-aimed a body.

Eductor Selection Table by Duty

The table condenses the five parameters into starting recommendations by duty: every row a starting point, confirmed against the vendor’s curves at your actual pressure and viscosity.

Duty Typical vessel / flow Starting motive flow Motive pressure Amplification ratio Body material Layout notes
Storage tank blending 10–50 m³ 15–40 m³/h 2–3 bar 4:1–5:1 316SS or PP One body per 2–3 m of length, floor-aimed
Plating / rinse tank agitation 1–5 m³, corrosive 2–10 m³/h 1.5–3 bar 3:1–4:1 PVC or PP Mini eductors, keep ports clear of drag-out
pH / reagent equalization 5–30 m³ 5–25 m³/h 2–4 bar 4:1 316SS or PVC Aim along floor to erase probe dead zones
Solids suspension 10–100 m³, dirty liquor 25–80 m³/h 3–4.5 bar 3:1–4:1 316SS + hard nozzle insert Highest turnover band; replaceable insert
Sump / pit jet pumping Lift 3–6 m Motive 2–4× lift flow 4–7 bar 1:1–2:1 CS or 316SS External pump, no submerged rotating parts
Chemical dilution / dosing Inline or day tank 1–10 m³/h 2–5 bar 1:1–4:1 PVC / PVDF / PTFE Suction line from concentrate drum
Vacuum generation Filtration / degassing Small motive flow 3–6 bar n/a (gas suction) 316SS Operates 3–6 kPa abs; different throat sizing

The bottom rows remind that an eductor is a general jet-pump device, not only a tank mixer: dilution and vacuum use the same Venturi geometry in different regimes, with the same five parameters applied to a different duty. That is the whole method in one line.

Eductor vs Mechanical Pump: The Cost Comparison

The economic case for an eductor loop is not that the eductor is cheap. It is that the pump, motor, electrics and maintenance schedule are sized for a fraction of the flow. What matters is whole-system cost over several years.

Take the earlier 20 m³ tank at four turnovers per hour: an 80 m³/h circulation requirement. The eductor route needs 16–20 m³/h of motive flow at 2–3 bar: a 2.2–3 kW pump class, small pipe, standard electrical feed. The mechanical route needs an 80 m³/h circulation pump at whatever head the loop demands, call it 5.5–7.5 kW of installed power once motor efficiency and loop losses are counted, plus bigger pipe, valves and electrics. The agitator alternative is a 5.5–7.5 kW geared drive with a shaft seal through the tank wall. Hydraulic power follows the same arithmetic: the 20 m³/h motive stream at 2.5 bar is roughly 1.4 kW of fluid power, against roughly 4.4 kW for the 80 m³/h stream at 2 bar, before pump and motor losses.

Over a five-year life, the differences compound through three cost layers. Capital: the eductor loop buys a small pump and a handful of molded bodies; the mechanical routes buy a big pump or a gearbox, a bigger motor and heavier pipe, typically several times the first cost. Energy: the eductor loop pays for the motive stream only, a quarter to a fifth of the circulation flow, at continuous running and 0.10–0.15 USD/kWh, a four-figure annual difference on a duty like this. Maintenance: the eductor loop has one wearing item, the nozzle insert, replaceable in minutes, against pump seals, impeller wear and motor bearings, or an agitator’s seal, bearings and gearbox. The maintenance story in one sentence: an eductor system has no rotating parts in the tank.

The honest caveats keep the comparison fair. An eductor loop reaches uniformity more slowly than a big agitator churning the whole volume at once, it trades peak power for sustained, gentle circulation, and duties that genuinely need high shear are not eductor duties. And if the pump you need for the motive stream does not already exist, part of the capital saving disappears. The standard check before any purchase: what pump is on site, what does it deliver at the tank, and does that sit inside an eductor’s window? If yes, the marginal cost of adding eductor mixing is close to the cost of the eductor itself.

Common Eductor Selection Mistakes

The failures below recur across commissioning reports; each maps to a parameter in this guide.

  • Sizing from pump nameplate pressure instead of pressure at the eductor. Line losses and elevation drop 1–2 bar off a long motive run, pushing the body below its minimum and killing suction. Measure at the body.
  • Specifying the best-case amplification ratio. Designing at 5:1 and running at 3:1 means the turnover spec is never met, and the only fix is buying more bodies. Size at the worst case.
  • Ignoring discharge back-pressure. Tank level rises, head on the discharge rises, entrainment falls. The same body behaves differently full and nearly empty.
  • Under-submerging the body. Ports pull air at low level and the ratio collapses, then the system is blamed instead of the level control.
  • Wasting the plume. Discharge aimed upward churns the top and leaves the bottom stratified. Aim along the floor or long axis.
  • One body for a long tank. A single eductor cannot clear the far corners of a 6+ meter vessel; multiple bodies on a header, aimed to reinforce each other, is the layout that works.
  • Oversizing the manifold the wrong way. An undersized header starves the far bodies; an oversized one costs money but works. Size for uniform pressure and vent the header.
  • Ignoring solids. Abrasive liquor wears the nozzle bore first; without a replaceable insert, the amplification ratio decays silently until someone measures it.
  • Forgetting what the eductor is for. Specifying an eductor for a 200-meter transfer line or high-shear emulsification fails not because the eductor is bad but because it is the wrong device class, and that decision comes before the math.

Each mistake shares one root cause: selecting from the catalog’s best numbers instead of the installation’s real ones. Run the five parameters against the worst case and most of this list never happens.

Eductor Selection Checklist

  • Duty confirmed as circulation, blending, suspension, lift or dilution, not long-distance transfer
  • Tank volume and required turnovers per hour written down before any catalog opens
  • Circulation requirement divided by a conservative amplification ratio (3:1–4:1) for motive flow
  • Pump curve read at the eductor, not the pump flange; pressure inside the body’s window with margin
  • Discharge back-pressure checked at the worst-case tank level
  • Amplification ratio confirmed at actual viscosity and submergence
  • Body material matched to fluid, with a replaceable nozzle insert for abrasive service
  • Submergence below operating low level; suction ports above the settled solids layer
  • Discharge aimed along the floor / long axis; bodies spaced one per 2–3 m
  • Header sized for uniform pressure, purged and vented
  • Turnover and pressure logged at commissioning as the diagnosis baseline

If you have a tank and a pump and want to know whether an eductor loop is the answer, send the vessel volume, the liquid and the pump curve to the engineering desk and the sizing can be confirmed from your numbers. For the support behind the table, the eductor product range covers the body types, and two guides go deeper into this territory: what is an eductor for the Venturi principle and entrainment physics, and mixing a tank with a small pump for the induction-ratio arithmetic and placement rules in detail. When a loop is already installed and underperforming, eductor troubleshooting works backwards from the symptoms to the cause.

Next Step

Send the Duty. Get Sized Nozzles Back.

Send your flow, pressure, fluid and target coverage. We come back with nozzle options and figures, not a catalogue number.

RC

Written by

Ray Chan

Industrial spray nozzle specialist. I size tank cleaning, atomizing, flat-fan and spiral nozzles against real duty conditions, flow, pressure, fluid and target, rather than catalogue numbers. Every guide here comes from actual sizing work.

← Back to Guides