CPCE Computational Fluid Dynamics Theory & Principles 4 — Questions and Answers
Question 1: The PISO (Pressure Implicit with Splitting of Operators) algorithm differs from SIMPLE primarily in that it:
- Solves the momentum equation only once per time step
- Performs multiple pressure corrections per time step without inner iterations (Correct answer)
- Uses an explicit time advancement for the pressure field
- Solves pressure on a staggered grid
Correct answer: Performs multiple pressure corrections per time step without inner iterations
PISO performs a predictor and two (or more) corrector steps per time step, making it more suitable for transient flows than SIMPLE's iterative approach.
Question 2: Which type of mesh is most suitable for complex geometries due to its ability to conform to curved boundaries?
- Structured Cartesian mesh
- Unstructured mesh with triangles/tetrahedra (Correct answer)
- Overlapping (Chimera/overset) mesh
- Uniform rectilinear mesh
Correct answer: Unstructured mesh with triangles/tetrahedra
Unstructured meshes with triangular or tetrahedral elements can conform to arbitrarily complex geometries, making them ideal for industrial CFD applications.
Question 3: The 'checkerboard' pressure instability in co-located grid arrangements arises because:
- The time step is too large for the chosen mesh
- Pressure and velocity are evaluated at the same locations, decoupling alternating pressure modes (Correct answer)
- The viscosity is computed using incorrect gradients
- Boundary conditions are not properly imposed
Correct answer: Pressure and velocity are evaluated at the same locations, decoupling alternating pressure modes
On co-located grids, standard central differencing for the pressure gradient does not sense a checkerboard pressure field, allowing spurious oscillatory modes to persist.
Question 4: In Reynolds-Averaged Navier-Stokes (RANS) modeling, the Reynolds stress tensor arises from:
- Time-averaging the nonlinear convective term, yielding correlations of velocity fluctuations (Correct answer)
- Spatial filtering of the pressure gradient
- Linearization of the viscous stress tensor
- Ensemble averaging of the boundary conditions
Correct answer: Time-averaging the nonlinear convective term, yielding correlations of velocity fluctuations
Time-averaging the nonlinear u·∇u term produces the Reynolds stress tensor −ρ⟨u'u'⟩, which must be modeled to close the RANS equations.
Question 5: The wall-y⁺ parameter is used to determine whether:
- The mesh is fine enough to resolve the viscous sublayer or if wall functions should be used (Correct answer)
- The turbulence model transition criterion is satisfied
- The CFL condition is met near solid walls
- The pressure is correctly prescribed at no-slip walls
Correct answer: The mesh is fine enough to resolve the viscous sublayer or if wall functions should be used
y⁺ = u_τ·y/ν indicates the location of the first mesh point in wall units; y⁺ < 5 implies the viscous sublayer is resolved, while y⁺ ≈ 30–300 suits wall functions.
Question 6: Which finite volume concept ensures that the flux leaving one cell face exactly equals the flux entering the adjacent cell?
- Consistency
- Conservativeness (Correct answer)
- Stability
- Boundedness
Correct answer: Conservativeness
Conservativeness is the property that guarantees flux conservation at cell interfaces, which is a fundamental advantage of the finite volume method.
Question 7: The Boussinesq eddy-viscosity hypothesis assumes that Reynolds stresses are:
- Proportional to the square of the turbulent kinetic energy
- Proportional to the mean rate-of-strain tensor via a scalar turbulent viscosity (Correct answer)
- Independent of the mean flow gradients
- Equal to the molecular viscous stresses
Correct answer: Proportional to the mean rate-of-strain tensor via a scalar turbulent viscosity
The Boussinesq hypothesis expresses Reynolds stresses as μ_t times the mean strain rate, analogous to how viscous stresses relate to strain in Newtonian fluids.
The PISO (Pressure Implicit with Splitting of Operators) algorithm differs from SIMPLE primarily in that it: