Hydraulic

Gravity-Flow Rate Calculation Between Two Reservoirs

Q: Why do I need to calculate the full and empty reservoir cases?

Because the static head ΔZ that drives the liquid decreases as the reservoir empties. With the tank full, ΔZ is at its maximum and the flow rate is the highest possible; with the tank empty, ΔZ is at its minimum and the flow rate is the lowest. The design must meet the required flow rate in the worst case (usually empty) and respect the velocity limit in the best case (full).

Q: How is the flow rate found if the loss depends on the flow rate itself?

The problem is implicit: head loss grows with v² (that is, with Q²), while the available energy ΔZ is fixed in each case. The operating flow rate is the one at which the total loss exactly equals ΔZ. The calculator solves this numerically by bisection, adjusting Q until h_dist + h_loc = ΔZ to a precision of 10⁻¹⁰ m of head.

Q: Can I use Hazen-Williams instead of Darcy-Weisbach?

Hazen-Williams is simpler, but it is only valid for water at ambient temperature and within a restricted velocity range. Darcy-Weisbach with the Colebrook-White friction factor is universal — it holds for any fluid, temperature, and flow regime — which is why it is the method adopted here. For rigorous industrial design, prefer Darcy-Weisbach.

Q: What happens if the calculated flow rate is lower than required?

It means the available head does not overcome the losses for the target flow rate. The remedies are: increase the diameter (strong effect, since h_dist ∝ 1/D⁵), reduce fittings and length, raise the reservoir (increase ΔZ), or, if none of that is enough, add a pump — at which point it is no longer a pure gravity line.

Q: Does atmospheric pressure affect the calculation?

Not directly, as long as both ends are open to the atmosphere (free surfaces). In that case the pressure terms cancel in the Bernoulli balance and only the static head and the losses remain. If the receiving tank is pressurized, you must add the pressure difference (converted to m of head) to the loss side.

Q: Why doesn't the flow rate double when I double the head?

Because head loss grows with the square of the flow rate. Since ΔZ ≈ k·Q², the flow rate is proportional to the square root of the head (Q ∝ √ΔZ). Doubling ΔZ increases the flow rate by only about 41%, not 100%.

In a gravity line there is no pump: the driver of the flow is the static head difference (ΔZ) between the free liquid surface in the reservoir and the discharge point. The flow rate settles when all of that available energy is consumed by the head loss of the pipeline.

When to use

Use this method whenever you need to transfer liquid from an elevated reservoir to a lower consumer or tank without a pump — process feed from an elevated surge tank, condensate return, tank drainage, transfer between vessels in plants with favorable elevation, or overflow lines. Because the reservoir empties over the course of operation, the available head varies: a correct check requires evaluating both extreme cases (full reservoir = maximum ΔZ = maximum flow rate; empty reservoir = minimum ΔZ = minimum flow rate) to ensure the line delivers the required flow range throughout the entire campaign.

What gravity flow is

A gravity line transfers liquid from a higher point to a lower one without any hydraulic machine. The sole driver of the motion is the static head difference (ΔZ) between the free liquid surface in the reservoir and the discharge point. All of that potential energy is spent overcoming pipe friction; when the expenditure equals the available energy, the flow rate settles at an equilibrium value.

It is one of the most economical and reliable fluid-transport arrangements — no energy consumption, no moving parts, no pump maintenance — but also the least flexible: the flow rate is imposed by the physics of the system, not chosen by the operator. Sizing it well means ensuring the delivered flow range serves the process from the start to the end of operation.

Theoretical basis

The starting point is the Bernoulli equation applied between the two free surfaces. When both are open to the atmosphere and the level changes slowly (negligible velocity in the tanks), the pressure and kinetic-energy terms at the ends cancel, and the balance reduces to:

ΔZ = h_dist + h_loc

In other words: all the available head turns into head loss. Unlike pump sizing — where you compute the total head the machine must supply for a given flow rate — here the flow rate is the unknown, and ΔZ is the fixed input.

The difficulty lies in the non-linearity: the head loss, both friction (Darcy-Weisbach) and minor (ΣK), depends on velocity squared, and velocity depends on the flow rate itself. There is no closed-form solution — the problem is solved iteratively.

How the method works, step by step

Estimate a flow rate Q and convert it to velocity: v = (Q/3600)/(π·D²/4).
Compute the Reynolds number Re = ρ·v·D/μ to identify the regime.
Obtain the friction factor f — laminar (f = 64/Re) for Re < 2000, or the Serghides estimator for Re ≥ 4000, with cubic interpolation in the transition zone.
Sum the losses: h_dist = f·(L/D)·v²/2g plus h_loc = ΣK·v²/2g.
Compare with ΔZ: if the total loss is less than ΔZ, there is energy to spare and the flow rate can rise; if it is greater, the flow rate must fall.
Bisection: the algorithm adjusts Q repeatedly until h_dist + h_loc = ΔZ to a precision of 10⁻¹⁰ m of head. That Q is the operating flow rate.

The procedure is run twice — once with the full-tank ΔZ (maximum flow rate) and once with the empty-tank ΔZ (minimum flow rate) — producing the complete operating range of the line.

Practical design considerations

Diameter is the dominant lever. Since h_dist ∝ 1/D⁵, a single size larger drastically reduces the loss and greatly raises the flow rate. It is almost always the most effective adjustment when the line does not deliver enough.
Velocity within the economic range (≈ 0.6 to 2.0 m/s). Very high velocities cause noise, erosion, and water hammer; very low ones promote sedimentation. Check the velocity in the full case, where it is highest.
The empty case usually fails the design. It is the worst case for flow rate. If the consumer requires a minimum, it is this case — not the full one — that must satisfy it.
Do not neglect the fittings. In short runs packed with bends, h_loc can exceed h_dist. Include the tank entrance (K ≈ 0.5) and the exit into the receiver (K ≈ 1.0).
Mind the Q ∝ √ΔZ relationship. When the loss is dominated by the v² term, doubling the head increases the flow rate by only ~41%. This explains why the flow rate falls relatively little between full and empty in well-sized systems.

Link to standards and methods

The calculation combines the Darcy-Weisbach equation (friction loss) with the Colebrook-White equation solved by the Serghides estimator. The fitting K coefficients follow established tables (Hydraulic Institute / manufacturer manuals), and the approach is consistent with the head-loss principles of ABNT NBR 5626 for pipe flow. By using Darcy-Weisbach instead of restricted empirical formulas (such as Hazen-Williams), the method remains valid for any fluid, temperature, and flow regime — a requirement of rigorous industrial design.

Formulas and fundamentals

Energy balance (Bernoulli) for the gravity line ΔZ = h_dist + h_loc

Form of the Bernoulli equation between the two free surfaces, both at atmospheric pressure and with negligible velocity. ΔZ (m) is the static head difference between the reservoir surface and the discharge point; h_dist (m) is the friction (major) head loss in the pipe; h_loc (m) is the sum of the minor losses. All the potential energy turns into head loss — there is no pump head.

Friction head loss (Darcy-Weisbach) h_dist = f · (L/D) · (v² / (2·g))

f is the friction factor (dimensionless); L the run length (m); D the inside diameter (m); v the mean velocity (m/s); g = 9.81 m/s². The dependence on v² makes the loss strongly non-linear with flow rate, which is why the solution is iterative.

Friction factor (Colebrook-White (Serghides), turbulent regime) 1/√f = −2·log₁₀( ε/(3.7·D) + 2.51/(Re·√f) )

Explicit Serghides approximation of Colebrook-White for Re ≥ 4000. ε is the absolute wall roughness (m); Re = ρ·v·D/μ is the Reynolds number. Below 2000 the laminar form (f = 64/Re) applies, and in the transition zone (2000–4000) a cubic interpolation is used.

Minor (local) loss from fittings h_loc = (Σ K) · (v² / (2·g))

ΣK is the sum of the loss coefficients of all fittings (bends, valves, entrance/exit, tees). Because h_dist and h_loc depend on the same v², the total loss grows with the square of the flow rate.

Velocity from flow rate v = Q / A = (Q/3600) / (π·D²/4)

Converts the flow rate Q (m³/h) into velocity v (m/s) over the cross-sectional area A (m²). It is the bridge between the search variable (Q) and the loss terms, which depend on v and Re.

Standards & methods

Darcy-Weisbach equation (friction head loss)
Colebrook-White equation (Serghides estimator)
Colebrook-White friction-factor method (Serghides estimator)
ABNT NBR 5626 (building water systems — flow and head loss)
Hydraulic Institute Engineering Data Book (fitting K coefficients)

Typical reference values

Quantity	Typical range	Note
Economic velocity in gravity lines	0.6 to 2.0 m/s	Above ~3 m/s noise and erosion increase; below ~0.6 m/s there is a risk of sedimentation.
Absolute roughness ε — commercial carbon steel	0.045 to 0.10 mm	PVC/HDPE: ~0.0015–0.007 mm; rusted steel: 0.15–0.5 mm.
Reynolds number — turbulent threshold	Re ≥ 4000	Re < 2000 is laminar; 2000–4000 is the unstable transition zone, to be avoided in design.
K coefficient — long-radius 90° elbow	0.2 to 0.4	Tank outlet (sharp-edged entrance): K ≈ 0.5; discharge into a tank: K ≈ 1.0.
Typical friction factor f (water, commercial steel)	0.018 to 0.030	Drops slowly as Re rises; increases with relative roughness ε/D.

Worked example

Gravity water transfer between two tanks

Inputs

ΔZ tank full: 8.0 m
ΔZ tank empty: 4.0 m
Line length (L): 60 m
Inside diameter (D): 100 mm
Roughness (ε): 0.046 mm
Σ K (fittings): 6.5 —

Results

Maximum flow rate (full): ≈ 85 m³/h
Minimum flow rate (empty): ≈ 60 m³/h
Velocity (full): ≈ 3.0 m/s
Reynolds (full): ≈ 3.0×10⁵ —
Friction factor f: ≈ 0.018 —

The line delivers between 60 and 85 m³/h as the reservoir empties — a ~30% drop in flow rate between full and empty, consistent with the relationship Q ∝ √ΔZ when the loss is dominated by the v² term (halving ΔZ from 8 to 4 m multiplies the flow rate by ≈ 1/√2). The velocity of 3.0 m/s in the full case is normal and acceptable for water in a steel line — at the upper edge of the economic range but with no erosion concern — and it drops to ≈ 2.1 m/s once the tank is empty. The design at D = 100 mm is reasonable. The binding question is the minimum flow rate: if the consumer needs more than ≈ 60 m³/h, it is the empty case — not the full one — that governs, and the diameter would have to grow (h_dist ∝ 1/D⁵) to lift the empty-tank flow rate.

Run your own calculation Create a free account and get this for your own inputs in seconds.

Common mistakes

Checking only the full-reservoir case: it is the minimum flow rate (empty tank) that determines whether the line still serves the consumer at the end of the batch.
Confusing ΔZ with the height of the pipe column — what matters is the elevation difference between the free liquid surface and the discharge point, not the length of the pipe.
Neglecting minor losses: in short lines with many bends and valves, ΣK·v²/2g can exceed the friction loss and pull down the real flow rate.
Adopting a fixed friction factor (e.g., 0.02) instead of computing f as a function of Re and ε/D — the flow-rate error can exceed 10% in small diameters.
Forgetting the exit loss into the receiving tank (K ≈ 1.0), which dissipates all the remaining kinetic energy.
Undersizing the diameter: since h_dist varies with 1/D⁵, dropping one pipe size can cut the flow rate in half.

Frequently asked questions

Why do I need to calculate the full and empty reservoir cases?

Because the static head ΔZ that drives the liquid decreases as the reservoir empties. With the tank full, ΔZ is at its maximum and the flow rate is the highest possible; with the tank empty, ΔZ is at its minimum and the flow rate is the lowest. The design must meet the required flow rate in the worst case (usually empty) and respect the velocity limit in the best case (full).

How is the flow rate found if the loss depends on the flow rate itself?

The problem is implicit: head loss grows with v² (that is, with Q²), while the available energy ΔZ is fixed in each case. The operating flow rate is the one at which the total loss exactly equals ΔZ. The calculator solves this numerically by bisection, adjusting Q until h_dist + h_loc = ΔZ to a precision of 10⁻¹⁰ m of head.

Can I use Hazen-Williams instead of Darcy-Weisbach?

Hazen-Williams is simpler, but it is only valid for water at ambient temperature and within a restricted velocity range. Darcy-Weisbach with the Colebrook-White friction factor is universal — it holds for any fluid, temperature, and flow regime — which is why it is the method adopted here. For rigorous industrial design, prefer Darcy-Weisbach.

What happens if the calculated flow rate is lower than required?

It means the available head does not overcome the losses for the target flow rate. The remedies are: increase the diameter (strong effect, since h_dist ∝ 1/D⁵), reduce fittings and length, raise the reservoir (increase ΔZ), or, if none of that is enough, add a pump — at which point it is no longer a pure gravity line.

Does atmospheric pressure affect the calculation?

Not directly, as long as both ends are open to the atmosphere (free surfaces). In that case the pressure terms cancel in the Bernoulli balance and only the static head and the losses remain. If the receiving tank is pressurized, you must add the pressure difference (converted to m of head) to the loss side.

Why doesn't the flow rate double when I double the head?

Because head loss grows with the square of the flow rate. Since ΔZ ≈ k·Q², the flow rate is proportional to the square root of the head (Q ∝ √ΔZ). Doubling ΔZ increases the flow rate by only about 41%, not 100%.

Glossary

ΔZ (static head difference): Elevation difference between the free liquid surface in the reservoir and the discharge point. It is the potential energy per unit weight available to drive the flow, expressed in meters of liquid column (m of head).
Friction head loss (h_dist): Energy dissipated by friction along the pipe wall, calculated with Darcy-Weisbach. It grows with length, with velocity squared, and falls sharply as the diameter increases.
Minor (local) head loss (h_loc): Energy dissipated in fittings and singularities (bends, valves, entrances, exits), expressed as h_loc = ΣK·v²/2g. In short lines it can dominate the total loss.
Friction factor (f): Dimensionless Darcy-Weisbach coefficient that quantifies flow friction. It depends on the Reynolds number and the relative roughness ε/D; computed via Colebrook-White (Serghides estimator).
Reynolds number (Re): Ratio of inertial to viscous forces (Re = ρ·v·D/μ). It defines the regime: laminar (Re < 2000), transition (2000–4000), or turbulent (Re > 4000).
Full / empty case: The two extreme states of the reservoir. Full gives the maximum ΔZ and maximum flow rate; empty gives the minimum ΔZ and minimum flow rate — together they bound the operating range of the line.