CFD PAST 5 CFU EXAMS
EXAM 1
Rezolvare:
Ex 1:
a) Discretized Equation (implicit, 1st-order in time, 2nd-order in space)
b) Mesh Quality Indexes and Impacted Terms (2 pt)
Key mesh quality indicators:
Orthogonality → affects diffusion term (since Laplacian depends on angle align-
ment)
Aspect ratio → affects accuracy of both advection and diffusion
Skewness → impacts gradient and Laplacian terms
Non-uniformity → introduces interpolation errors
Most affected terms:
Diffusion term: very sensitive to non-orthogonal meshes
Advection term: highly impacted by non-uniform and skewed meshes
The most influential mesh quality parameters are orthogonality, skewness, aspect ra-
tio, and smoothness of the mesh. The diffusion term is particularly sensitive to non-
orthogonality, as this introduces errors in the calculation of gradients used in the
Laplacian. Skewness affects both the advection and diffusion terms by reducing the
accuracy of interpolation. The advection term, being hyperbolic, is most affected by
mesh alignment with the flow direction, and therefore both aspect ratio and skewness
have a
critical influence. Poor mesh quality can lead to numerical diffusion, instability, and
loss of accuracy, especially for convective-dominated problems.
c) Moving mesh case
When using a moving mesh strategy and assuming the diffusion coefficient µ=0\mu =
0µ=0, the advection-dominated transport equation must be adapted to account for
the mesh motion. This is typically done using an Arbitrary Lagrangian-Eulerian (ALE)
fluid and the mesh, such that U r⃗ el=U ⃗−w ⃗\vec{U}_{rel} = \vec{U} - \vec{w}Urel=U−w,
formulation. The convective term is modified to use the relative velocity between the
where w ⃗\vec{w}w is the mesh velocity. The equation becomes ∂ /
∂t+∇⋅( (U ⃗−w )⃗ )=q\partial \phi/\partial t + \nabla \cdot (\phi(\vec{U} - \vec{w})) =
q∂ /∂t+∇⋅( (U−w))=q, ensuring that the fluxes are correctly evaluated in the context
of a deforming control volume.
d) Source Term Discretization Strategy 1(pt)
For better convergence:
Linearize the source term: write q=SP P+SUq = S_P \phi_P + S_Uq=SP P+SU
This allows the source to be treated in the matrix form, improving diagonal domi-
nance and stability
Avoid strong nonlinear source coupling unless iteratively handled
To improve the convergence of the solution, the source term should be discretized
using a linearization strategy. This involves expressing the source term as a function
of the dependent variable, typically in the form q=SP P+SUq = S_P \phi_P +
S_Uq=SP P+SU, where SPS_PSP is the linear coefficient and SUS_USU is the con-
stant part. This approach integrates the source term directly into the matrix coeffi-
cients during assembly, improving diagonal dominance and enabling faster and more
stable convergence of the iterative solver.
Ex 2:
The SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) algorithm for solv-
ing the incompressible Navier-Stokes equations begins by guessing a pressure field.
The momentum equations are solved using this guessed pressure to obtain a provi-
sional velocity field that does not satisfy continuity. A pressure correction equation is
then derived from the continuity equation and used to update both the pressure and
velocity fields. The corrected fields are used in the next iteration, and this process
continues until convergence. The algorithm effectively decouples pressure and veloci-
ty while maintaining mass conservation, which is essential for incompressible flows.
b)
To adapt segregated solvers originally designed for incompressible flows to handle
compressible flows, modifications must be made to account for density variations.
This includes solving the energy equation in addition to the mass and momentum
equations and introducing a suitable equation of state that relates pressure, tempera-
ture, and density. The solver must also allow for pressure waves and acoustic effects,
which are not present in incompressible flow simulations. Therefore, a pressure-based
coupled solver or a density-based solver may be used depending on the Mach num-
ber regime.
c)
To solve the compressible flow equations, it is necessary to assign well-posed bound-
ary conditions based on the type of boundary and the direction of flow (inflow or out-
flow). At the inlet, one may prescribe total pressure, total temperature, and flow direc-
tion if the flow is subsonic, or all flow variables for supersonic inflow. At the outlet, sta-
tic pressure is commonly specified for subsonic conditions, while no conditions are
imposed for supersonic outflows. At solid walls, a no-slip condition is imposed for ve-
locity, and either adiabatic or fixed wall temperature is set for energy. For symmetry or
axisymmetric boundaries, zero gradient conditions are used. Ensuring that the num-
ber of boundary conditions matches the number of characteristics entering or leaving
the domain is essential for mathematical well-posedness
d)
When simulating flows with phase transition, multiple classes of non-equilibrium can
arise. These include mechanical non-equilibrium, where the pressure between phases
differs; thermal non-equilibrium, where the phases have different temperatures; and
chemical non-equilibrium, particularly in reacting flows or mass transfer situations.
The speed of sound in a two-phase mixture is not uniquely defined under non-equilib-
rium conditions. In the homogeneous equilibrium model (HEM), a mixture speed of
sound can be derived based on averaged properties. However, under non-equilibrium,
additional modeling is required, and there is no universally accepted definition. The
speed of sound may depend on the relaxation times between phases and the type of
wave considered (e.g., acoustic, shock, or interfacial waves)
EXAM 2
Ex 1:
Ex 2:
EXAM 3
Ex 1:
b)
Ex 2:
EXAM 4 ( 09.06.2025 )
Exercise 1:
a) Explain the momentum conservation in 2d
b) Explain what is discretization with 1st order time implicit and 2nd order space
c) Explain how could you enforce the matrix to be diagonaly dominant.
Exercise 2:
a) Introduce the formulation of the Boussiesq hypothesis applied to the momentum
conservation in 2D
b) Give example of two models with wall functions, their applications and mathemati-
cal differences
c) Explain the steps for point b) in simple method
d)Explain the steps for point b) under relaxation
e) Explain which solver is the best to use in this case according to the given answers
in a), b), c), d)
Ex1:
Ex2: