Conservation of Mass (Continuity) in Problem Solving
1. The Master Integral Continuity Equation
In classical fluid mechanics, matter is neither created nor destroyed. For any closed material fluid system $V_f(t)$, the rate of change of mass is identically zero:
Setting the intensive property $b = 1$ in the universal Reynolds Transport Theorem (Eq. 3.8) yields the Master Integral Mass Conservation Equation for any moving control volume $V_c(t)$:
For a fixed, rigid control volume ($V_0, \vec{v}_c = \vec{0}$), this simplifies to the Eulerian storage and convective flux balance:
Mass Conservation: Complete Derivation Chain (Notes.pdf, Eqs. 3.9-3.11 and 4.4-4.7)
Step 1: Mass balance for an arbitrary moving control volume (Eq. 3.10)
The principle (Eq. 3.9) states that the mass of a fluid volume does not change, $dM/dt=0$ with $M=\int_{V_f(t)}\rho\,dV$. Choose $\phi=\rho$ in the Reynolds Transport Theorem (Eq. 3.8) and replace the left-hand side by zero:
Moving the flux term to the other side shows the physical meaning: the rate of change of the mass stored in the control volume equals minus the net mass flux out through its boundary, that is, the mass flux entering.
Step 2: Fixed control volume (Eq. 3.11)
For a fixed volume $V_0$ with surface $\Sigma_0$ we have $\vec{v}_c=\vec{0}$ and the domain does not depend on $t$. The limits of integration are therefore constants, no boundary term appears, and the time derivative passes inside the integral (compare Eq. 3.7 with $\vec{v}_c=\vec{0}$):
Substituting into Eq. 3.10 with $\vec{v}_c=\vec{0}$ gives
Step 3: Mass flow rate and sign convention
With $\vec{n}$ outward, the algebraic mass flow rate through a part $A$ of the control surface is
It is positive at outlets, negative at inlets and zero on impermeable walls, where no fluid crosses the surface and $(\vec{v}-\vec{v}_c)\cdot\vec{n}=0$. Eq. 3.10 then reads $dM_c/dt=-\sum_A\dot m_A$. For a steady flow in a fixed volume the left-hand side vanishes and $\sum\dot m_{\text{out}}=\sum\dot m_{\text{in}}$ (magnitudes). On a port of area $A_k$ with uniform $\rho_k$ and uniform velocity $v_k$ normal to the port, $|\dot m_k|=\rho_kv_kA_k$.
Step 4: Differential form (Eqs. 4.4-4.7 of Notes.pdf, Chapter 4)
The following steps belong to Chapter 4 of the Notes and are reproduced here to close the logical chain between the integral and the local statement. Apply Eq. 3.6 with $\phi=\rho$ to Eq. 3.9:
Gauss' theorem (Eq. 2.9), $\int_\Sigma\vec{A}\cdot\vec{n}\,d\sigma=\int_V\nabla\cdot\vec{A}\,dV$ with $\vec{A}=\rho\vec{v}$, turns the surface integral into a volume integral over the same domain:
Arbitrary-volume argument. Let $g(\vec{x},t)=\partial\rho/\partial t+\nabla\cdot(\rho\vec{v})$, a continuous function. Eq. 4.4 holds for every fluid volume. Suppose $g(\vec{x}^{*},t)\neq0$ at some point, say $g(\vec{x}^{*},t)\gt0$. By continuity $g\gt0$ in a ball $B_\varepsilon(\vec{x}^{*})$. The fluid volume that occupies this ball at time $t$ then gives $\int_{B_\varepsilon}g\,dV\gt0$, which contradicts Eq. 4.4. The case $g\lt0$ is identical. Hence $g\equiv0$ at every point:
Expand the divergence, $\nabla\cdot(\rho\vec{v})=\vec{v}\cdot\nabla\rho+\rho\,\nabla\cdot\vec{v}$, and use the material derivative $D\rho/Dt=\partial\rho/\partial t+\vec{v}\cdot\nabla\rho$:
For a constant-density fluid $D\rho/Dt=0$, so $\nabla\cdot\vec{v}=0$ (Eq. 4.7). For a steady flow of a gas, $\partial\rho/\partial t=0$ and Eq. 4.5 gives $\nabla\cdot(\rho\vec{v})=0$ (Eq. 4.8).
Step 5: Control volume bounded by a moving free surface (extension)
Extension, not in the Notes. For a tank, the free surface $z=h(t)$ is part of $\Sigma_c$ and moves with $\vec{v}_c=\dot h\,\vec{e}_z$. It is a material surface (no fluid crosses it), so the kinematic condition $\vec{v}\cdot\vec{n}=\vec{v}_c\cdot\vec{n}$ holds. Therefore $(\vec{v}-\vec{v}_c)\cdot\vec{n}=0$ on it and it contributes no flux to Eq. 3.10. Only the inlet and the outlet contribute. With constant density and a tank of constant horizontal area $A_T$, the stored mass is $\rho A_Th(t)$ and Eq. 3.10 gives
The discharge law $Q_{\text{out}}=A_e\sqrt{2gh}$ (Torricelli) is an input of the problem. It follows from Bernoulli's equation for steady inviscid flow, which is not derived in Topic 3 of the Notes, so it is taken here as an assumption.
2. Incompressible Fluids & 1D Streamtubes
When the fluid density is spatially and temporally constant ($\rho = \text{const}$, valid for liquids and for gas flows where the Mach number $Ma = v/a < 0.3$), the density factors out of both integrals. For a fixed control volume:
where $Q = \int_A (\vec{v}\cdot\vec{n}) dA$ is the volumetric flow rate ($\text{m}^3/\text{s}$).
1D Streamtube Approximation & Average Velocity
Across any duct or streamtube cross-section $A$, the mass flow rate $\dot{m}$ ($\text{kg/s}$) is:
where the area-averaged bulk velocity $\bar{v}$ is defined by:
In laminar circular pipe flow (Poiseuille), $v(r) = v_{\max}[1 - (r/R)^2]$, so the bulk velocity is $\bar{v} = \frac{1}{2}v_{\max}$. In turbulent pipe flow, the profile is fuller ($1/7\text{th}$ power law), yielding $\bar{v} \approx 0.817 v_{\max}$. Always integrate the profile explicitly when given!
3. Unsteady Storage: Filling & Emptying Tanks
When the mass inside the control volume changes with time, the unsteady storage term must be evaluated by tracking the moving liquid level $h(t)$.
Consider a tank of horizontal cross-sectional area $A_T$ with an inlet flow rate $Q_{\text{in}}$ and a bottom discharge orifice of area $A_e \ll A_T$. Defining a control volume that encloses the liquid inside the tank up to the free surface $z = h(t)$:
Net mass flux across the boundaries:
Applying Torricelli's law for inviscid discharge from the bottom orifice, $v_e(t) = \sqrt{2gh(t)}$, so $Q_{\text{out}} = A_e \sqrt{2gh(t)}$. Substituting into the continuity equation:
For a draining tank with no inflow ($Q_{\text{in}} = 0$) and initial height $H_0$, separating variables and integrating yields the classical draining time:
Airfoil Wake Deficit: Determining Upstream Capture Width
Physical Setup: An airfoil is tested in a uniform wind tunnel flow of velocity $v_\infty$ and density $\rho$. Far downstream, boundary layer skin friction and wake mixing generate a symmetric velocity deficit over a wake band of width $2b$ ($|y| \le b$):
where $0 < \Delta < 1$ is the centerline fractional velocity defect. Far upstream where the flow is completely uniform ($v = v_\infty$), all the fluid that eventually exits through the wake $[-b, b]$ passed through a capture band $[-y_{-\infty}, y_{-\infty}]$.
Objective: Determine the upstream capture semi-width $y_{-\infty}$ from conservation of mass.
Define a 2D control volume bounded by:
- Upstream inlet face ($\Sigma_{\text{in}}$): Located at $x = -\infty$, extending across $y \in [-y_{-\infty}, y_{-\infty}]$ where velocity is uniform $v_x = v_\infty$.
- Downstream exit face ($\Sigma_{\text{out}}$): Located at $x = +\infty$, spanning the exact wake thickness $y \in [-b, b]$ with velocity profile $v(y)$.
- Upper and lower boundaries ($\Sigma_{\text{streamline}}$): The limiting streamlines that pass through $(\pm b)$ at downstream infinity. By definition, no fluid crosses a streamline: $(\vec{v}\cdot\vec{n}) \equiv 0$.
For unit span ($W = 1$) into the page, integrating over the upstream aperture:
Exploiting symmetry between upper and lower halves of the wake:
Evaluating the spatial integral across the shear layer:
Therefore, the total wake mass flow rate is:
Since the flow is steady and no mass escapes through the bounding streamlines, $\dot{m}_{\text{in}} = \dot{m}_{\text{out}}$:
Dividing by $2\rho v_\infty$ gives the exact analytical result:
If you choose a rigid rectangular control volume of fixed height $2H > 2b$ instead of streamline boundaries, fluid must leak outward through the top and bottom surfaces with mass rate $\dot{m}_{\text{lateral}} = 2\rho v_\infty (b - y_{-\infty}) = \rho v_\infty b \Delta$. Failing to include this lateral mass escape is the most frequent exam mistake in wake momentum balances!