Dimensional Analysis & Similitude - Theory & Concepts

Dimensional Analysis & Similitude

$Learning Objectives$

Express hydraulic variables consistently in MLT and FLT dimensional systems and test equations for dimensional homogeneity.
Apply Buckingham Pi using dimensional rank, independent repeating variables, and a physically complete variable set.
Interpret Reynolds, Froude, Euler, Weber, Mach, Strouhal, and cavitation parameters without overstating what any one number represents.
Distinguish geometric, kinematic, and dynamic similarity and recognize that complete dynamic similarity may require several dimensionless groups.
Derive model-prototype velocity, time, discharge, pressure, force, and power scale laws from a declared scale convention.
Apply Froude and Reynolds similarity with explicit density, viscosity, and gravity assumptions.
Diagnose incompatible similarity requirements and quantify the direction of important scale effects.
Analyze distorted models and derive scale relations from separate horizontal and vertical scales.
Extend similitude thinking to numerical models through mesh, time-step, turbulence-model, and boundary-condition verification.

$Why Dimensional Analysis Is More Than Unit Checking$

Dimensional analysis constrains the mathematical form of a physical relationship and reveals the independent nondimensional parameters that govern similarity. Similitude then uses those parameters to decide what a model can reproduce faithfully, what must be distorted, and what uncertainty remains when exact similarity is impossible.

$Dimension$

A statement of the physical nature of a quantity in terms of independent fundamental quantities such as mass, length, and time.

$Unit$

A chosen numerical standard used to measure a dimensional quantity, such as metres, feet, seconds, or newtons.

$Common MLT Dimensions$

Quantity	Symbol	MLT dimensions
Length	$L$	$L$
Time	$t$	$T$
Velocity	$V$	$LT^{-1}$
Acceleration	$a$	$LT^{-2}$
Mass density	$\rho$	$ML^{-3}$
Dynamic viscosity	$\mu$	$ML^{-1}T^{-1}$
Kinematic viscosity	$\nu$	$L^2T^{-1}$
Pressure or stress	$p$	$ML^{-1}T^{-2}$
Force	$F$	$MLT^{-2}$
Surface tension	$\sigma$	$MT^{-2}$
Power	$P$	$ML^2T^{-3}$

$MLT and FLT Systems Must Not Be Mixed Mid-Derivation$

The MLT system treats mass, length, and time as fundamental. The FLT system treats force, length, and time as fundamental, with mass derived from $M=FT^2/L$ . Either system works, but all variables in one exponent calculation must use the same fundamental basis.

$Dimensional Homogeneity$

The requirement that all additive terms in a physically meaningful equation have identical dimensions and that both sides of an equation have the same dimensions.

$Dimensional Consistency Check$

STEP-BY-STEP

Replace every dimensional variable by its fundamental dimensions.
Reduce the exponents of each independent fundamental dimension.
Confirm that every additive term has the same dimensions.
Confirm that arguments of logarithmic, exponential, trigonometric, and similar transcendental functions are dimensionless.
Check empirical unit-system constants separately; dimensional homogeneity alone does not validate their numerical value.

$Dimensionally Homogeneous Does Not Mean Physically Correct$

Both $V=\sqrt{gH}$ and $V=100\sqrt{gH}$ are dimensionally homogeneous. Dimensional analysis can reject impossible dimensional forms, but theory or experiment is still required to determine numerical coefficients, signs, functional form, and physical applicability.

$Buckingham Pi Theorem$

If a physically complete relation contains $n$ dimensional variables whose dimensional matrix has rank $r$ , the relation can be expressed using $n-r$ independent dimensionless groups.

Buckingham Pi Representation

Reduction of a dimensional relation to independent dimensionless groups.

F(q_1,q_2,\ldots,q_n)=0 \quad\Longrightarrow\quad \Phi(\Pi_1,\Pi_2,\ldots,\Pi_{n-r})=0

$Use Dimensional Rank, Not a Casual Count of Symbols$

The number $r$ is the rank of the independent fundamental dimensions represented by the variable set. Merely seeing $M$ , $L$ , and $T$ somewhere in a table does not guarantee $r=3$ if the chosen variables do not span those dimensions independently.

$Buckingham Pi Method$

STEP-BY-STEP

Define the physical question and list every variable that can materially influence the response.
Express all variables in one fundamental dimensional system.
Determine the dimensional rank $r$ .
Select $r$ repeating variables that collectively span the fundamental dimensions and cannot form a dimensionless product among themselves.
Multiply each nonrepeating variable by the repeating variables raised to unknown powers.
Set the net exponent of every fundamental dimension to zero and solve the exponent system.
Check that the resulting Pi groups are independent.
Rewrite groups into physically recognizable forms when useful; equivalent powers or products of valid groups are also valid.
Test limiting cases and compare with theory or data before interpreting the reduced relation.

Buckingham Pi reduction

Variables are reduced into independent dimensionless groups.

$An Omitted Physical Variable Cannot Be Recovered Later$

Buckingham Pi can reorganize the variables supplied to it, but it cannot discover a variable that was never included. If roughness, gravity, surface tension, compressibility, geometry ratio, frequency, cavitation pressure, or another effect matters physically, it must appear in the original variable set.

$Dimensionless Groups Are Not Unique$

If $\Pi_1$ and $\Pi_2$ are independent dimensionless groups, then combinations such as $\Pi_1^2$ , $1/\Pi_1$ , or $\Pi_1\Pi_2$ may also be dimensionless. The goal is not to reproduce one memorized algebraic arrangement; it is to obtain an independent set that exposes the controlling physics clearly.

Rough-Pipe Pressure-Loss Similarity Form

One useful nondimensional representation for rough internal flow.

\frac{\Delta p}{\rho V^2} =\Phi\left( \frac{\rho VD}{\mu}, \frac{\varepsilon}{D}, \frac{L}{D} \right)

$Physical Meaning of the Rough-Pipe Groups$

The left side is an Euler-type pressure coefficient. The right-side groups are Reynolds number, relative roughness, and relative length. Dimensional analysis identifies the governing form; an empirical or theoretical closure such as a friction-factor relation is still needed to calculate the pressure drop.

$Geometric Similarity$

Similarity in which all corresponding lengths have constant scale ratios and corresponding angles and relevant shape ratios are preserved.

$Kinematic Similarity$

Similarity in which corresponding velocity vectors have consistent scale ratios and directions, producing similar streamline and motion patterns.

$Dynamic Similarity$

Similarity in which the relevant ratios of physical forces or equivalent controlling dimensionless groups are matched between model and prototype.

$Matching One Convenient Parameter Is Not Complete Dynamic Similarity$

Equal friction factor, equal Reynolds number, or equal Froude number alone establishes only the similarity associated with the physics represented by that parameter. Complete dynamic similarity requires every materially important independent Pi group to be matched or its mismatch to be shown negligible for the response of interest.

$Scale Convention Used in This Lesson$

All scale ratios below use prototype divided by model:

\lambda=\frac{L_p}{L_m}

A $1:20$ model therefore has $\lambda=20$ . Other references sometimes use the reciprocal. Always write the definition of the scale ratio before applying an exponent.

Model-to-prototype scaling

Corresponding quantities are related through declared scale ratios.

$Reynolds Number$

A dimensionless parameter measuring the relative importance of inertial and viscous effects.

Reynolds Number

Inertia-viscosity similarity parameter.

Re=\frac{\rho VL}{\mu}=\frac{VL}{\nu}

$Froude Number$

A dimensionless velocity ratio comparing flow speed with the gravity-wave speed associated with the characteristic length.

Froude Number

Gravity-inertia similarity parameter for free-surface and gravity-driven flow.

Fr=\frac{V}{\sqrt{gL}}

$Froude Number Is Not Literally the Force Ratio$

With the usual scaling, $Fr^2=V^2/(gL)$ corresponds to an inertia-to-gravity force-scale ratio. The Froude number $Fr$ is its square root and is best interpreted as a characteristic velocity divided by a gravity-wave velocity scale.

$Euler Number$

A dimensionless parameter comparing a pressure difference with inertial dynamic pressure.

Euler Number

Pressure-inertia similarity parameter.

Eu=\frac{\Delta p}{\rho V^2}

$Weber Number$

A dimensionless parameter comparing inertial effects with surface-tension effects.

Weber Number

Inertia-surface-tension similarity parameter.

We=\frac{\rho V^2L}{\sigma}

$Mach Number$

A dimensionless velocity ratio comparing flow speed with the acoustic wave speed of the medium.

Mach Number

Compressibility similarity parameter based on acoustic speed.

Ma=\frac{V}{c}

$Strouhal Number$

A dimensionless frequency parameter comparing a characteristic oscillation time scale with the convective time scale.

Strouhal Number

Unsteadiness or periodic-flow similarity parameter.

St=\frac{fL}{V}

$Cavitation Number$

A dimensionless pressure-margin parameter comparing absolute pressure above vapor pressure with dynamic pressure.

Cavitation Number

Pressure margin above vapor pressure normalized by dynamic pressure.

\sigma_c=\frac{p_{\text{abs}}-p_v}{\tfrac12\rho V^2}

$Selecting the Governing Similarity Requirement$

Use Froude similarity when gravity and inertia govern a free surface, such as spillways, rivers, hydraulic jumps, and ship waves.
Use Reynolds similarity when viscosity controls boundary layers, friction, wakes, separation, or internal-flow structure.
Use Weber similarity when surface tension affects jets, bubbles, droplet breakup, aeration, or very small waves.
Use cavitation-number similarity when vapor formation and collapse affect the response.
Use Mach or related compressibility similarity when density change or pressure-wave propagation matters.
Use Strouhal similarity when oscillation frequency or transient timing is part of the prototype question.

Selecting a governing group

Dominant physics identifies the appropriate similarity parameter.

$Choose Similarity from the Prototype Physics, Not the Laboratory Equipment$

A model is not valid because it fits the flume, pump, or test rig. First identify the dominant forces and response quantities in the prototype, then choose a model scale and fluid that make the relevant dimensionless groups as faithful as practical.

Froude Similarity Scale Laws

Prototype-to-model ratios for equal Froude number and equal gravitational acceleration.

\frac{V_p}{V_m}=\lambda^{1/2}

\frac{t_p}{t_m}=\lambda^{1/2}

\frac{Q_p}{Q_m}=\lambda^{5/2}

Froude Pressure, Force, and Power Scales

General density-ratio form for equal gravity under Froude similarity.

\frac{\Delta p_p}{\Delta p_m}=\frac{\rho_p}{\rho_m}\lambda

\frac{F_p}{F_m}=\frac{\rho_p}{\rho_m}\lambda^3

\frac{P_p}{P_m}=\frac{\rho_p}{\rho_m}\lambda^{7/2}

$Same-Fluid Froude Models$

When model and prototype use the same liquid, $\rho_p/\rho_m=1$ , so pressure scales as $\lambda$ , force as $\lambda^3$ , and power as $\lambda^{7/2}$ . These exponents follow from the declared prototype/model convention and must be inverted if a reference defines scale the other way around.

General Reynolds Similarity Scale Laws

Prototype-to-model scaling with explicit kinematic-viscosity ratio.

\frac{V_p}{V_m}=\frac{\nu_p/\nu_m}{\lambda}

\frac{Q_p}{Q_m}=\lambda\left(\frac{\nu_p}{\nu_m}\right)

\frac{t_p}{t_m}=\frac{\lambda^2}{\nu_p/\nu_m}

$Same-Fluid Reynolds Models Can Demand Extreme Model Velocity$

If $\nu_p=\nu_m$ , then $V_p/V_m=1/\lambda$ , or $V_m=\lambda V_p$ . A small water model may therefore require very high model velocity, pressure, and power and may encounter cavitation or free-surface effects that were not dominant in the prototype.

$Why Froude and Reynolds Similarity Usually Conflict$

For the same fluid and gravity, Froude similarity requires $V_p/V_m=\lambda^{1/2}$ , while Reynolds similarity requires $V_p/V_m=1/\lambda$ . Both can hold simultaneously only for $\lambda=1$ . Reduced free-surface models therefore usually prioritize Froude similarity and then quantify the unmatched Reynolds-number effect.

Froude-Reynolds similarity conflict

Same-fluid geometric models generally cannot match both gravity and viscous similarity.

$Quantify the Unmatched Group Instead of Calling It Negligible$

After choosing the dominant similarity law, calculate important unmatched groups in both model and prototype. A statement such as “viscous effects are small” should be supported by Reynolds-number magnitude, sensitivity tests, multi-scale experiments, calibration, or other evidence appropriate to the problem.

$Explore Model-Prototype Scaling$

Use the interactive calculator to compare Froude and Reynolds scaling under the same prototype/model convention and to examine how changing fluid viscosity alters the required velocity ratio.

Hydraulic Model Scaling Simulator

Learning objective: Compare model and prototype scaling and see why the dominant Froude or Reynolds requirement governs different hydraulic experiments.

Model : prototype = 1 : $\lambda$ , where $\lambda=L_p/L_m$ . All displayed ratios are prototype divided by model.

Governing similarity law

Geometric scale1 : 10

Model velocity1.00 m/s

Model discharge0.10 m³/s

Prototype prediction

3.162 m/s

velocity

31.623 m³/s

discharge

Governing equality: $Fr_m=Fr_p$
$V_p/V_m$: 3.16228
$Q_p/Q_m$: 316.22777
$t_p/t_m$: 3.16228

V_p/V_m=\lambda^{1/2}

Q_p/Q_m=\lambda^{5/2}

Froude similarity matches inertia and gravity. Reynolds, Weber, and cavitation effects may remain distorted at small model scales.

$Scale Effect$

A systematic model-prototype difference caused by one or more relevant dimensionless groups, geometric details, material properties, or measurement limitations not being reproduced at the same nondimensional value.

$Common Hydraulic Scale Effects$

Viscous influence becoming too large in a small Froude-scaled model.
Surface tension distorting shallow waves, jets, bubble breakup, or air entrainment.
Geometric roughness becoming impossible to reproduce because prototype roughness does not scale below available material grain size.
Boundary-layer thickness and separation changing with Reynolds number.
Cavitation appearing too early or being suppressed because absolute pressure similarity is not maintained.
Air bubbles, sediment grains, leakage, and instrument resolution failing to scale geometrically.

$Managing Scale Effects$

STEP-BY-STEP

Identify the response quantities the model must predict.
Rank the dimensionless groups that influence those responses.
Match the dominant group exactly where practical.
Calculate every important unmatched group in model and prototype.
Choose a sufficiently large model, adjusted fluid property, pressure level, temperature, or roughness treatment where safe and practical.
Test more than one scale or operating point when scale-effect trends are uncertain.
Calibrate and validate against prototype or field data when available.
Report uncertainty and the range over which extrapolation is credible.

$Distorted Model$

A physical model that intentionally uses different scale ratios for different geometric directions or selected dimensions.

Slope Scaling in a Vertically Distorted Model

Prototype/model slope relation for horizontal scale lambda_H and vertical scale lambda_V.

\frac{S_p}{S_m}=\frac{\lambda_V}{\lambda_H}

\frac{S_m}{S_p}=\frac{\lambda_H}{\lambda_V}

Variables

Symbol	Description	Unit
$\lambda_H$	Prototype-to-model horizontal length scale	-
$\lambda_V$	Prototype-to-model vertical length scale	-

$Consequences of Geometric Distortion$

Vertical exaggeration changes slope, curvature, hydraulic radius, wave propagation, sediment behavior, secondary circulation, and sometimes the controlling nondimensional groups. Once horizontal and vertical scales differ, standard undistorted Froude exponents cannot simply be copied; the required scale relations must be re-derived for the distorted geometry.

$Hydraulic Roughness Does Not Scale by Length Alone$

Even if physical roughness height is reduced geometrically, its hydraulic effect depends on relative roughness and Reynolds number. Small models often need artificial roughness or calibration to reproduce prototype resistance, especially in rivers, floodplains, and turbulent conduits.

$Similitude in Numerical Models$

Computational models remove the need for a laboratory length scale but do not remove scale reasoning. Grid spacing, time step, numerical diffusion, turbulence closure, wall treatment, interface resolution, and boundary conditions can change the effective nondimensional physics. Mesh independence, time-step independence, conservation checks, and validation remain numerical counterparts of physical-model similitude.

$Hydraulic Model Quality Checks$

State the prototype question and the response quantities to be predicted.
List all physically relevant variables before forming dimensionless groups.
Declare the scale convention and every fluid-property ratio used.
Identify the dominant similarity law and justify it physically.
Calculate important unmatched dimensionless groups in model and prototype.
Check roughness, cavitation, surface tension, air entrainment, sediment, and instrumentation scale effects where relevant.
Preserve boundary conditions, control geometry, and operating range that materially affect the response.
Perform uncertainty analysis and, where possible, tests at multiple scales or operating points.
Validate against theory, limiting cases, and prototype or field observations before extrapolation.

$Hydraulic Similitude Workflow$

STEP-BY-STEP

Define the prototype decision or prediction the model must support.
Assemble a physically complete variable set and determine the governing Pi groups.
Rank gravity, viscosity, pressure, surface tension, compressibility, unsteadiness, roughness, and other relevant effects.
Declare the prototype/model geometric scale convention.
Choose the dominant similarity requirement and derive all dependent scale ratios from it.
Compute unmatched groups and identify the likely direction and magnitude of scale effects.
Select model size, fluid, pressure, roughness, instrumentation, and operating range to control those effects.
Validate the model before using it for prototype extrapolation.
Report assumptions, scale effects, calibration, and uncertainty with the final prediction.

Key Takeaways

Dimensional homogeneity is necessary for a physical equation but cannot prove that the equation is physically correct.
Buckingham Pi uses a physically complete variable set and dimensional rank to produce $n-r$ independent nondimensional groups.
Dimensionless groups are not algebraically unique; independence and physical meaning matter more than memorizing one arrangement.
Froude number is a characteristic velocity ratio; $Fr^2$ corresponds to the usual inertia/gravity force-scale ratio.
Complete dynamic similarity requires all materially important independent groups to be matched or their mismatch to be justified as negligible.
Every scale law depends on the declared scale convention; this lesson uses prototype/model ratio $\lambda=L_p/L_m$ .
Froude scaling governs gravity-driven free-surface similarity, while Reynolds scaling governs inertia-viscosity similarity and explicitly depends on viscosity ratio.
Reduced models usually cannot match every relevant group, so scale effects must be quantified, managed, validated, and reported.
Distorted physical models and discretized numerical models both require re-derived scaling logic and independent verification.

Previous TopicHydraulic Machinery - Examples & Applications

Quiz Me

Next TopicDimensional Analysis & Similitude - Examples & Applications

Prev Next

Quiz Me