Jawaharlal Nehru Technological University Hyderabad B.Tech III COMPUTATIONAL AERODYNAMICS Previous Question Paper (August - 2022 - R18)

The continuity equation represents the principle of conservation of mass. For a control volume moving with the fluid, the mass of the fluid element remains constant over time. Considering a fluid element of volume V with density ρ, the mass m is given by the integral of ρ dV over the volume. Since the mass is constant, the material derivative of the mass is zero:

Applying the Reynolds Transport Theorem to this integral, we obtain the integral form of the continuity equation for a moving fluid element:

Since this holds for any arbitrary control volume, the integrand must be zero, leading to the differential form of the continuity equation:

This can be expanded as:

The transformation between models is defined by the choice of control volume:

• Fixed Control Volume (Eulerian approach): The integral form is ∫_CV ∂ρ/∂t dV + ∫_CS ρ(V · n) dA = 0, where the first term represents the rate of change of mass inside the fixed volume and the second term represents the net flux across the control surface.
• Moving Control Volume (Lagrangian approach): The mass is tracked by following the fluid particles, resulting in the material derivative form Dρ/Dt + ρ(∇ · V) = 0.
• Incompressible Flow: For fluids where density is constant (Dρ/Dt = 0), the equation simplifies to the divergence-free condition: ∇ · V = 0.

A diagram would depict a small fluid element moving along a streamline, showing the mass flux entering through one face and exiting through the opposite face, accounting for the change in density and velocity across the element dimensions.

In computational fluid dynamics, the classification of partial differential equations (PDEs) dictates how information propagates through the computational domain. The domain of dependence represents the region in the space-time domain that influences the solution at a specific point, while the range of influence describes the set of points that are affected by the solution at a specific point.

For hyperbolic equations, such as the wave equation or unsteady inviscid flow equations, information propagates at finite speeds. The domain of dependence is defined by the characteristic lines originating from a point in the past. If a point P is at (x, t), its domain of dependence is the interval [x-at, x+at] on the initial data line at t=0. Conversely, the range of influence is the region bounded by the characteristics emanating forward in time from P. In CFD, this implies that numerical schemes must satisfy the CFL (Courant-Friedrichs-Lewy) condition, ensuring the numerical domain of dependence encompasses the physical domain of dependence to maintain stability.

For elliptic equations, such as the Laplace or Poisson equations representing steady-state potential flows, the domain of dependence for any point is the entire closed domain. Information propagates at infinite speed, meaning a disturbance at any single point instantaneously affects the solution everywhere in the domain. Consequently, elliptic problems require boundary conditions to be specified on all boundaries of the closed region simultaneously.

The impact on CFD is fundamental: hyperbolic problems are solved using time-marching methods where the solution evolves based on local information, requiring careful selection of time steps to respect the characteristic cones. Elliptic problems are solved using relaxation or direct matrix inversion techniques because the global nature of the solution necessitates simultaneous convergence across the entire grid. Failure to respect these propagation characteristics leads to numerical instability in hyperbolic solvers or non-convergence in elliptic solvers.