Computing Flow Over A Cylinder

Volumetric Geometry/Domain Preparation   |   Pressure based & Steady   |   Turbulence (k-omega SST)   |   Data Post-Processing

Minor Project - Flow over Cylinder,
Flowthermolab
05/09/2025

Project Overview

This project presents a steady-state CFD analysis of a Flow over a cylinder to investigate:

• Under-Relaxation Factor and Pseudo Transient approaches
• Compare Steady State and Transient Solver, monitor velcoity and pressure at a point in wake region, and perform FFT Temporal variation of pressure and velocity
• Compare Pressure and Density Based Solvers

Problem Statement

Compute th efollowing cases over the cylinder in 2D:

Cases	Problem Statement
Case 1	Simulate steady state flow over a cylinder in water medium and Compare Under-Relxation factor and pseudo transient approachs
Case 2	Simulate transient problem: Visualize the Flow capturign wake fomation monitor velocity and pressure at point in wake Perfrom FFT on temporal variation of velocity and pressure
Case 3	Simulate steady state flow over a cylinder and compare the pressure and density based solvers

Software Used

1.   Ansys Space-Claim ------- Geometry Cleanup and refinement
2.   Ansys Fluent Mesher ----- Mesh / Grid Generation
3.   Ansys Fluent ------------- Computation
4.   Ansys Post Processor ---- Post Processing
5.   MATLAB ------------------ Data Analysis & Scientific Plots

Methodology

1. Geometry Cleanup and refinement

• Geometry Dimensions
• Diameter = 1000 mm
• Flow Field:
• Length = 25000 mm
• Width = 20000 mm

Domain prepared (Mixing Elbow)

2. Mesh Details

Properties	Mesh Details	Values
Sizing	Use Adpative sizing Adaptive Resolution Edge Sizing Element Size	Yes 7 0.025 m 0.1 m
Quality	Target Skewness Smoothing	Yes High
Automatic method	Sheet Body Method Sweepable Body Method	Quad Dominant Sweep
Inflation	Inflation Option Maximum layers Boundary	Smooth Transition 24 Wall Cylinder
Face Sizing	Element Size Behaviour	70 mm Hard

• Mesh Statistics:
  • Nodes: 53,944
  • Elements: 53,507

Flow Field Mesh & Cylinder Wall Inflation

3. Solver Settings

Solver Settings were changed for different cases and every case had more number of compuation with different boundary condition settings involved to get better results and data to conclude with some recommendation

4. Results

Case 1. Velocity Contours

Velocity Contours(Flow Doamin XY plane)

Case 1. Pressure Distribution (Cylinder Wall)

Pressure Distribution (Cylinder Wall)

Case 2: Velocity and Pressure Contours (Transient and Steady State)

Velocity Contours and Pressure for Transient and Steady State

Case 2: Velocity Streamlines (Transient Case)

Velocity Streamlines Transient Case

Case 2: Velocity and Pressure Profile at a point (Transient Case)

Here the Velocity and Pressure profiles for cylinder were monitored in wake region at a point having coordinates as follow:

• (1, 0.5, 0)

Velocity and Pressure Profile at a point in wake region of the cylinder

Case 2: FFT Spectral Analysis of Drag Coefficient Report (Transient Case)

At point 5 (1, 0.5, 0). X-axis Range was clipped from 885270 to 1020000.

• Y-axis Variable: Coefficient of Drag, Y-axis Function: Magnitude
• X-axis Variable: Flow time, X-axis Function: Frequency [Hz]

FFT Spectral Analysis of Drag Coefficient Report

Force	Steady Case	Transient Case	Difference (%)
Drag	1.48437e-06 [N]	6.1913e-06 [N]	76
Lift	3.72491e-10 [N]	3.5189e-11 [N]	5.85

Case 3: Velocity Contours for DBS & Coupled PBS

Velocity Contours for DBS & Coupled PBS

Case 3: Velocity Profile capturing Forebody wake and Transition

• For Forebody wake: A line was created in the wake region of the cylinder where the velocity data was plotted. (1,2,0) & (1,-2,0)
• For Transition: A line was created in the wake region of the cylinder where the velocity data was plotted. (1,0.7,0) & (-1,0.7,0)

Note: The Velocity was calcuated on points plotted on the yellow line shown in the figure beside the plot.

Velocity Profile capturing Forebody wake and Transition

Case 3: Coefficient of Drag

• Unrealistic High Cd Values for SIMPLE, SIMPLEC and PISO
• For the Presssure Based Solver (Coupled), Density Based Solver (Implicit, Implicit_AUSM, Explicit) average Cd value is 0.680756616

Case	Drag	Cd
1: PBS: SIMPLE	2.12499e+10 [N]	847114.304
2: PBS: SIMPLEC	1.2502e+10 [N]	498384.6055
3: PBS: PISO	3.30817e+10 [N]	1318781.795
4: PBS: Coupled	14020.4 [N]	0.558914696
5: DBS: Implicit	12311.6 [N]	0.490794426
6:DBS: Implicit (AUSM)	7471.94 [N]	0.297864331
7:DBS: Explicit	34503.3 [N]	1.375453012

Discussion

For Case 1:
  Four Different cases for under relaxation factor and pseudo transient approach were run to compare and understand which method is better and why?
As being evident from the results Coupled Automatic flow is fully developed and achieved its convergence way better than other settings, coupled user specified flow seems to have converged in residuals, the values had drastic fall and then paralleled that is flow did not develop fully how ever for about 800 iterations there was no anomality seen to disturb any values. In Simple Default setting flow require more time to get better, also the residuals were observed to be unstable they were in control. For Simple User specified case, flow development was slow here the unstable observation was damped.
Conclusion can be made that; Coupled Pseudo transient approach is better than others.
To Remember: Carefully specify the value to coupled user specified settings to get better results. And results of SIMPLE Scheme can be improved by using Pseudo Time in the settings and there must be balance between time or unstable residual (What is preferred by the user)

For Case 2:

• For Cylinder, Steady and Transient cases were simulated with ~50k mesh elements and for velocity 0.00010048 m/s. It can be seen from both the cases residuals steady state seems to have converged faster while for transient case it took ~24hrs to generate the residuals and report as can be seen.
• And since to capture the wake study over time is essential thus in the results coefficient of drag has difference of ~76 [%] between the results of steady and transient states.
• It can be concluded from findings that transient cases must be simulated to capture the intricacies of the flow regimes, steady state can be used for preliminary study for mesh.

For Case 3:

• There were seven cases simulated for the cylindrical 2D Body of Diameter 1m, the purpose was to run compressible flow over the cylinder and study the outcomes of both pressure and density-based solvers.
• For the outcome first we have coefficient of drag, second stability, and third convergence. It was observed that the Density Based solvers are better in all three criteria, while in pressure-based solvers coupled scheme has been able to solve the compressible flow for 0.6 Mach. For other pressure-based solvers more than one simulations were run for different cases to delay convergence and increase stability at the run time, basically segregated pressure-based solvers (SIMPLE, SIMPLEC, and PISO) should be avoided to be used in compressible flows as they were able to predict the flow and did not meet the outcomes as coupled, implicit and explicit.
• For the Presssure Based Solver (Coupled), Density Based Solver (Implicit, Implicit_AUSM, Explicit) average Cd value is 0.680756616. Where the individual values are provided in the table on the previous slide.
• At last, it can be concluded that for Compressible flow following solvers work well for steady state:
  • Pressure Based Solver (Coupled Scheme),
  • Density Based Solvers (Implicit and Explicit Schemes).

Recommended Repository Structure

CFD-Flow-Over-Cylinder/
│
├── README.md
├── Geometry image
├── Mesh images
├── Fluent_Case_Files
├── Results & Data

Author:
Ansh Vishal,
Aerospace Engineer
anshvishal215@gmail.com
LinkedIn