The Mesh Convergence Index, more formally known as the Grid Convergence Index (GCI), is a

quantitative measure used in CFD to assess how much the solution changes as the
computational grid is refined. It provides a standardized way to report the uncertainty due to
mesh discretization and helps determine if your simulation results are independent of the grid
size.

How to Calculate the Grid Convergence Index (GCI)

1. Run simulations on at least three grids: coarse, medium, and fine, each with
increasing mesh resolution.

2. Choose a key output variable (e.g., resistance, lift, drag coe icient) and record its
value for each mesh.

3. Calculate the refinement ratio (rr), which is the ratio of grid spacings between two
successive meshes (e.g., medium/coarse, fine/medium).

4. Estimate the apparent order of accuracy (pp), typically using Richardson

extrapolation.

5. Compute the relative error (ϵϵ) between solutions on successive grids.

6. Apply the GCI formula:

GCI21=1.25⋅ϵ21r21p−1GCI21=r21p−11.25⋅ϵ21

where:

 ϵ21=S2−S1S1ϵ21=S1S2−S1 (di erence between medium and fine grid solutions)

 r21r21 is the grid refinement ratio between medium and fine grids

 pp is the observed order of accuracy

 1.25 is a commonly used safety factor

The same formula applies for the coarse-to-medium grid (GCI32GCI32).

7. Interpret the GCI: Multiply by 100 to express as a percentage. A lower GCI indicates
less uncertainty due to the mesh. For detailed studies, aim for a GCI of 1–5%; for
parametric studies, 5–10% is often acceptable.

Why Use GCI?

 Objectively quantifies mesh uncertainty and helps demonstrate that further mesh
refinement will not significantly change results.

 Supports reproducibility and transparency in CFD studies.

Example

If you simulate ship resistance on three meshes and find the resistance values stabilize as the
mesh is refined, a low GCI confirms mesh independence. If GCI remains high, further
refinement or mesh quality improvements are needed.
“The Grid Convergence Index (GCI) quantifies the error between simulation results through
di erent grid refinements, allowing you to assess the convergence level of a computational
model.”

This process is essential for ensuring the reliability and credibility of CFD results in ship hullform
analysis.

The numerical setup for CFD simulation of a ship involves defining a range of parameters that
govern the accuracy, stability, and physical fidelity of the simulation. Key parameters include:

 Hull Geometry

 3D CAD model of the ship hull (including appendages if necessary)

 Parametric definitions for hull modifications or optimization

 Mesh Parameters

 Mesh type (structured/unstructured)

 Cell size and refinement regions (especially near hull surface and free surface)

 Boundary layer mesh settings (prismatic layers for resolving near-wall flow)

 Total cell count

 Domain Setup

 Computational domain size (length, width, height)

 Placement of hull within the domain

 Boundary Conditions

 Inlet: velocity or flow rate specification

 Outlet: pressure or outflow condition

 Side and top boundaries: symmetry, slip, or wall conditions

 Bottom: wall or slip

 Hull surface: no-slip wall condition

 Fluid Properties

 Density and viscosity of water (and air if multiphase simulation)

 Surface tension (if free surface e ects are significant)

 Physical Models

 Turbulence model (e.g., k-ε, k-ω SST)

 Free surface model (e.g., Volume of Fluid – VOF)

  Gravity and reference direction

 Solver Settings

 Steady or unsteady simulation

 Discretization schemes (spatial and temporal)

 Convergence criteria (residuals, force convergence)

 Time step size (for transient simulations)

 Initial Conditions

 Initial velocity and pressure fields

 Initial water level (for free surface simulations)

 Ship Motion Constraints

 Fixed or allowed degrees of freedom (sinkage, trim, heave, pitch)

 Output and Monitoring

 Forces and moments on the hull

 Wave elevation and flow field data

 Residuals and convergence history

These parameters must be carefully defined and validated to ensure the CFD simulation
accurately represents the physical behavior of the ship in water and provides reliable results for
resistance, seakeeping, or optimization studies

Hull geometry parametrization plays a central role in the CFD setup for ship simulations,
particularly in design optimization, performance assessment, and sensitivity analysis. Its main
functions and benefits include:

 Defining the Design Space: Parametrization creates a systematic way to describe and
modify the hull shape using a set of geometric variables (e.g., beam, draft, bulbous bow
height, stern shape). This enables the creation of a “shape design space” where
di erent hull forms can be explored and compared e iciently.

 Enabling Automated Optimization: By expressing the hull shape through parameters,

optimization algorithms can automatically adjust these variables to minimize
resistance, improve seakeeping, or achieve other performance goals. Techniques like
Arbitrary Shape Deformation (ASD) allow the CFD process to search for optimal hull
forms by varying input parameters.

 Reducing Computational Cost: Instead of manually redesigning the hull for each test,
parametrization allows rapid generation of new hull variants, significantly speeding up
the simulation and optimization process.
  Supporting Sensitivity and Trend Analysis: Parametric studies reveal how changes in
specific hull features a ect performance metrics such as resistance, wave patterns,
and added resistance in waves. This helps designers understand which aspects of the
hull most influence hydrodynamic behavior.

 Improving Flexibility and Adaptability: Parametrization techniques can be adapted to

di erent vessel types and design requirements, making the CFD setup more flexible and
robust for a wide range of ship designs.

In summary, hull geometry parametrization is essential for e icient, systematic, and automated
exploration of hull designs in CFD, enabling both optimization and deeper understanding of the
relationship between hull shape and hydrodynamic performanc

Hull roughness is accounted for in CFD simulations of ship hull resistance primarily
through modifications to the wall-function approach in turbulence models, which
incorporate roughness functions to represent the e ect of surface texture on the boundary
layer flow. The main methods include:

 Modified Wall-Function Approach: The most common practice is to use a roughness

function (often based on sand-grain roughness) within the wall-function of the
turbulence model. This approach adjusts the velocity profile near the wall to account for
increased friction due to roughness, without explicitly modeling the roughness
geometry. The roughness height (e.g., equivalent sand-grain roughness) is specified as a
parameter in the CFD setup.

 Roughness Function Models: These models, such as the one based on Granville’s
similarity law, modify the logarithmic law of the wall to include the e ect of roughness.
This allows the simulation to predict the increase in skin friction and total resistance
due to rough surfaces, and has been validated against both flat plate and 3D hull
experimental data.

 Geometric Modeling of Roughness: In some advanced studies, the actual roughness

geometry is resolved directly in the mesh, but this is computationally expensive and
rarely used for full-ship simulations.

 Heterogeneous Roughness Distribution: CFD can also model the e ect of roughness
distributed unevenly along the hull (e.g., bow-rough, stern-rough), allowing analysis of
localized roughness impacts and supporting strategies like partial hull cleaning. Studies
show that roughness at the bow has the largest impact on resistance, followed by other
regions.

 Low-Reynolds Number Turbulence Models: These can be used with modified

boundary conditions to simulate roughness e ects, though this approach is less
common than wall-function modifications.

Summary Table
 Method Description Typical Use

Modified wall-function with Adjusts turbulence model near-wall treatment using Most common,
roughness roughness practical

Incorporates empirical/logarithmic corrections for

Roughness function models roughness Validated, widely used

Rare, research
Direct geometric modeling Resolves roughness shapes in mesh applications

Heterogeneous roughness Advanced, localized

distribution Models spatially varying roughness along hull studies

Validation: These approaches have shown good agreement with experimental data, particularly
for predicting the increase in frictional and total resistance due to hull roughness.

In practice, the modified wall-function approach with specified roughness height is the
standard method for accounting for hull roughness in ship resistance CFD simulations.

Mesh sensitivity and grid convergence are closely related concepts in CFD, both aimed at
ensuring the reliability and accuracy of simulation results by examining the influence of mesh
resolution. However, they have distinct emphases:

Term Definition & Focus Practical Use

Examines how solution variables change with mesh Used to assess spatial
refinement. The goal is to understand the influence of discretization errors and to
Mesh mesh size and quality on the results, identifying when determine if the mesh is su iciently
Sensitivity further refinement yields diminishing changes. fine for trustworthy results.

Refers to the process and confirmation that further mesh

refinement does not significantly change the results. This Confirms that the solution is
is often quantified using metrics like the Grid Convergence independent of the mesh, meaning
Grid Index (GCI) and is sometimes called a mesh independence any further refinement will not
Convergence study. materially a ect the outcome.

Key Points:

 Mesh sensitivity is the broader process of checking how sensitive your results are to
mesh changes.

 Grid convergence (or mesh independence) is achieved when results stabilize and
further mesh refinement causes negligible change—typically below a set tolerance (e.g.,
1%).
  Both require running simulations on at least two or three increasingly refined meshes
and comparing key outputs (e.g., resistance, pressure drop, velocity).

 Mesh sensitivity study is the necessary step to reach grid convergence, which is the
ultimate goal for reliable CFD results.

In summary, mesh sensitivity is the process; grid convergence is the condition you aim to
confirm through that process.