A siphon is an inverted U-shaped tube that moves liquid from a higher reservoir up over a barrier and down into a lower reservoir, driven entirely by the head difference between the two free surfaces. A typical 25 mm garden hose siphon with a 1.5 m head delivers around 30–40 L/min with no pump and no electricity. The mechanism solves the problem of emptying or transferring liquid across a wall or rim where gravity alone won't help. You see it everywhere from aquarium gravel cleaners to the Roman aqueduct inverted siphons at Aspendos.
Siphon Interactive Calculator
Vary tube diameter, head difference, and effective discharge coefficient to see siphon flow, velocity, pressure head, and losses.
Equation Used
The calculator applies Torricelli flow with an effective discharge coefficient: diameter sets tube area, head difference sets ideal velocity, and Cd reduces the result for real hose friction, bends, and entrance losses.
- Tube is fully primed with no air pocket at the crest.
- Discharge outlet is below the source free surface.
- Cd is an effective coefficient including entry, bend, and hose friction losses.
- Water density is approximated as 1000 kg/m3 and g = 9.81 m/s2.
Inside the Siphon
A siphon runs because the column of liquid on the long (downhill) leg weighs more than the column on the short (uphill) leg. Once the tube is full and both ends are submerged or the discharge end is below the source surface, that weight imbalance pulls liquid continuously over the crest. Atmospheric pressure on the source surface pushes liquid up the short leg to replace what falls out the long leg — but the actual driving force is the head difference, not the atmosphere. People get this wrong constantly. The atmosphere only sets the upper limit on how high the crest can sit above the source: roughly 10.3 m for cold water at sea level, and in practice you cap out around 7–8 m before dissolved gases come out of solution and break the column.
Priming is what makes or breaks a siphon. If you don't fill the tube completely before starting flow, you get an air pocket at the crest. That pocket expands as flow tries to develop, the column breaks, and the siphon stalls. You prime by submerging the whole tube, by sucking on the discharge end, or by using a one-way bulb pump. Once running, the siphon self-sustains until either the source drops below the inlet, air enters the inlet (vortex breakthrough — happens when submergence falls below about 2 × pipe diameter), or the discharge end rises above the source level.
What goes wrong in real builds is almost always one of three things. The crest is too high and you're operating near the cavitation limit, so any temperature rise gases out the column. The inlet is too close to the surface and a vortex sucks air in. Or the tubing has a low spot on the discharge side that traps air and chokes the cross-section. None of these are theoretical — they show up the first time you try to drain a 1000 L IBC tote with a 6 m piece of clear vinyl hose.
Key Components
- Source-side (short) leg: The intake tube that dips into the upper reservoir. Inlet must sit at least 2 × pipe diameter below the free surface to prevent vortex air entrainment. A coarse strainer keeps debris out — anything that blocks even 30% of the cross-section will stall flow.
- Crest (apex): The high point of the inverted U. Maximum theoretical height above the source is ~10.3 m for water at 20 °C and 1 atm; practical limit is 7–8 m before dissolved-air degassing breaks the column. Keep the crest as low as the obstacle allows.
- Discharge-side (long) leg: The downhill leg that does the actual work. Its outlet must be lower than the source surface — that vertical difference (Δh) is the head driving flow. Longer drop means higher velocity, capped by tube friction losses.
- Priming arrangement: Either submerge-and-cap, mouth suction, or a squeeze bulb on the discharge end. Without complete air evacuation the column breaks at the crest. Aquarium gravel siphons use a built-in shake-prime mechanism — a one-way flap that captures liquid on each pump stroke.
- Vacuum break / siphon break: An air-bleed hole near the crest used in plumbing codes (UPC requires one on dishwasher drains) to deliberately break the siphon and prevent backflow contamination. Hole diameter is typically 3–5 mm.
Who Uses the Siphon
Siphons show up wherever you need to move liquid across a barrier without electricity, or where a pump would be overkill. They handle anything from a 5 L fish tank to municipal-scale stormwater overflows. The trick is matching the head difference to the required flow rate and keeping the crest height honest.
- Agriculture: Furrow irrigation siphon tubes — 50 mm aluminum tubes laid over a head ditch bank to deliver measured flow to each furrow on cotton fields in California's Central Valley.
- Brewing: Auto-Siphon racking canes used by homebrewers and craft breweries like Russian River to transfer wort or beer from primary to secondary fermenter without disturbing the trub layer.
- Aquarium maintenance: Python No Spill Clean and Fill gravel vacuums — uses tap-water Venturi action to prime the main siphon, then gravity flow drains the tank.
- Emergency fuel transfer: Manual siphon pumps like the GasTapper or a basic squeeze-bulb hose for transferring diesel or petrol between vehicles or jerry cans, particularly during hurricane evacuations.
- Civil engineering: Inverted siphons under canals and roadways — the Hetch Hetchy aqueduct uses several inverted siphons to carry water across valleys to San Francisco.
- Spillway design: Siphon spillways at dams like the Glen Canyon auxiliary structures, which self-prime once reservoir level rises above the crest and pass large flows automatically.
The Formula Behind the Siphon
Siphon flow rate comes from Torricelli's law modified for tube friction. The formula gives you outlet velocity from the head difference between the source surface and the discharge outlet. At the low end of the typical operating range — say 0.3 m of head on a 25 mm hose — you get a trickle, useful for slow draining without disturbing sediment. At the high end, 3 m of head on the same hose, friction dominates and you no longer get the velocity the ideal equation predicts. The sweet spot for a domestic-scale 25 mm siphon sits around 1–2 m of head, where flow is brisk but friction losses stay below 30% of the ideal value.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| v | Outlet velocity (ideal, frictionless) | m/s | ft/s |
| Q | Volumetric flow rate | m³/s | ft³/s or gal/min |
| Δh | Head difference between source surface and discharge outlet | m | ft |
| g | Gravitational acceleration | 9.81 m/s² | 32.2 ft/s² |
| A | Internal cross-sectional area of the tube | m² | in² |
| Cd | Discharge coefficient accounting for friction and entry losses (typically 0.6–0.8 for plain hose) | dimensionless | dimensionless |
Worked Example: Siphon in draining a koi pond for liner replacement
You're draining a 12,000 L ornamental koi pond at a private estate near Victoria BC for liner replacement. The pond surface sits 1.4 m above the storm drain at the bottom of the garden. You're using 25 mm (1 inch) ID reinforced PVC hose, 8 m total length, routed over the rockery rim which sits 0.6 m above the pond surface. You need to know the flow rate you'll actually see, so you can plan how long the drain will take and whether to go grab a coffee or stand there with a net catching koi.
Given
- Δhnom = 1.4 m
- Dtube = 0.025 m
- Lhose = 8 m
- hcrest = 0.6 m above source
- Cd = 0.65 dimensionless
Solution
Step 1 — calculate the tube cross-sectional area:
Step 2 — at nominal 1.4 m head, compute ideal outlet velocity using Torricelli:
Step 3 — apply the discharge coefficient to account for hose friction and entry losses, then convert to flow rate:
That's brisk — the 12,000 L pond drains in roughly 2 hours, fast enough that you stay nearby but slow enough not to suck koi into the inlet if you've fitted a basic strainer.
Step 4 — at the low end of the typical operating range, with the pond half-drained the head drops to 0.7 m:
Flow has dropped by 30% even though the tank is only half-empty — that's the square-root relationship biting you. The last third of the pond will take noticeably longer than the first third.
Step 5 — at the high end, if you re-routed the discharge to a creek 3 m below the pond surface:
In theory. In practice the 8 m hose at that velocity sees friction losses pushing real Cd closer to 0.5, knocking actual flow back to ~115 L/min. Above about 2.5 m of head on this hose size, you'd switch to 38 mm hose to keep up.
Result
Nominal flow is approximately 100 L/min, draining the 12,000 L koi pond in just under 2 hours. To a bystander watching the discharge end, that's a steady stream about as thick as a thumb — clearly moving fast, no surging or gulping. Across the operating range the flow varies from 71 L/min near empty up to 147 L/min if you found a deeper discharge point, with the sweet spot for this hose size sitting around the original 1.4 m head. If you measure significantly less than 100 L/min — say 60 L/min — the most common causes are: (1) an air leak at the hose-to-strainer joint letting bubbles into the crest and partially breaking the column, (2) a partial blockage in the inlet strainer from algae or pond debris reducing effective area, or (3) the crest being higher than the measured 0.6 m because the hose sags upward at an unseen point over the rockery, raising the effective high point and increasing static lift losses.
Choosing the Siphon: Pros and Cons
Siphons have one job — moving liquid downhill across a barrier — and they do it without power. But there are alternatives, and the right choice depends on flow rate, lift required, and whether you can guarantee a continuous head difference.
| Property | Siphon | Centrifugal pump | Hand-cranked rotary pump |
|---|---|---|---|
| Power required | Zero — gravity only | 0.1–10 kW typical | Manual labour, ~50 W sustained |
| Flow rate (typical 25 mm line) | 30–150 L/min depending on head | 50–500 L/min | 5–20 L/min |
| Maximum lift above source | ~7–8 m practical (atmospheric limit) | ~7 m suction lift, unlimited discharge | ~6 m suction lift |
| Priming required | Yes — every start-up | Self-priming variants exist; otherwise yes | Yes, but built into the cranking action |
| Cost (small-scale install) | $10–50 for hose and strainer | $150–800 plus wiring | $80–250 |
| Reliability over a 10-year service life | No moving parts; fails only from air ingress or source depletion | Bearing/seal wear at 5,000–20,000 hours | Diaphragm or vane wear at 200–1,000 hours of cranking |
| Best fit | Continuous gravity transfer with available head | High flow, variable elevation, powered site | Low-flow remote sites, fuel transfer |
Frequently Asked Questions About Siphon
Almost always dissolved-gas breakout at the crest. As liquid sits in the high point, the local pressure drops below atmospheric — the higher your crest, the lower that pressure goes. Dissolved air comes out of solution and accumulates at the apex. Once the bubble grows large enough to span the tube cross-section, the liquid column breaks and flow stops.
The fix is to lower the crest height. Every 10 cm you drop the apex buys you measurable margin. If you can't lower the crest, fit a small bleed valve at the high point so you can purge accumulated gas without re-priming the whole line. Cold liquid holds more dissolved gas than warm, so this problem is worse on a hot afternoon than first thing in the morning.
No, and this trips up a lot of people. The 10.3 m limit isn't about how hard you can pull — it's the height of a water column that atmospheric pressure can support against gravity. Once the column is taller than that, the pressure at the top falls to the vapour pressure of water and the liquid boils, breaking the column. You can't beat it with a stronger pump because there's no liquid left at the top to pull on.
If you genuinely need to lift water more than ~8 m practical, you need a positive-displacement pump pushing from the bottom, not a siphon pulling from the top. Mercury can siphon over 76 cm because it's 13.6× denser, and theoretically a denser-than-water fluid would extend the limit proportionally.
The closed-form Torricelli equation ignores friction, which is fine for short tubes but dominates as length-to-diameter ratio (L/D) climbs above 200. An 8 m hose at 25 mm ID has L/D = 320, putting you firmly in the friction-limited regime. Roughly speaking, halving the hose ID at constant length cuts your flow rate by a factor of 5–6 because friction scales with the fifth power of diameter at turbulent flow.
Rule of thumb: if your measured flow is less than 60% of Torricelli's prediction, friction is your problem. Either go up one hose size (25 mm to 38 mm typically doubles flow) or shorten the run. For long runs on small head, you're often better off with a small pump than a siphon.
If the garden bed is genuinely lower than the tank outlet and you have continuous head, a siphon wins on every count — zero power, zero noise, zero maintenance, and you can leave it running unattended. A 25 mm siphon with 1 m head will deliver 80–100 L/min, plenty for soaker hoses or drip irrigation.
Switch to a pump only if (a) the destination is above the source, (b) you need flow rate above ~150 L/min, or (c) you need precise on/off control via a timer. The grey area is when you have only 0.2–0.3 m of head; siphon flow gets unreliably slow there and a small 12 V pump becomes more practical.
Head difference. The driving force is the vertical distance between the tank water surface and the bucket on the floor. Drop tank height by half and you don't quite halve the flow — you reduce it by √(0.5) ≈ 0.71 according to Torricelli, before friction is even considered. On a short tank, the discharge end might also be too close to vertical level with the source, especially if you're using a 5 gallon bucket that's tall.
Lower the bucket — put it on the floor instead of a chair, or use a longer hose run so the discharge end can sit lower. Going from a 30 cm to a 60 cm vertical drop will roughly double your flow rate, no other changes needed.
Siphon spillways on large dams routinely pass 50–500 m³/s, and proposed Lake Mead emergency siphons in the 2010s were sized for 600+ m³/s using multiple parallel pipes. The limit on a single barrel is the cavitation pressure at the crest combined with practical pipe diameter — at very large diameters, Reynolds numbers go enormous, friction becomes minor, but the crest pressure constraint stays the same.
Scaling up runs into a quieter problem: priming. A 2 m diameter siphon barrel holds tonnes of water at the crest, and you can't suck on the end of it. Large siphons use vacuum pumps, ejector primers, or self-priming geometry where rising flood water naturally fills and seals the crest as the reservoir rises.
References & Further Reading
- Wikipedia contributors. Siphon. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
