Case Study 1 — Sizing a Ventilation Duct: Volume Flow Rate as a Flux Integral

DataField.Dev

Case Study 1 — Sizing a Ventilation Duct: Volume Flow Rate as a Flux Integral

Field: Mechanical / HVAC engineering, fluid dynamics Calculus used: Flux of a velocity field (Sections 36.6–36.8), the graph and disk flux formulas (Section 36.7)

When a hospital commissions a new operating theater, one number on the engineer's spec sheet decides whether the room is safe: the air change rate, often twenty full room-volumes of fresh air every hour. To hit it, the supply duct must deliver a guaranteed volume of air per second. That quantity — cubic meters per second crossing a duct cross-section — is not a pressure, not a temperature, and not a velocity. It is a surface integral. Specifically, it is the flux of the air-velocity field through the cross-sectional surface,

$$Q = \iint_S \mathbf{v}\cdot d\mathbf{S},$$

and getting it right is the entire job. This case study follows one such calculation from an idealized textbook duct to the messy real profile an engineer actually measures, showing where every term of Section 36.6 earns its keep.

The Idealized Duct: Uniform Flow

Start with the cleanest possible model. Air moves down a circular duct of radius $a = 0.20$ m along the $z$-axis with a uniform velocity field $\mathbf{v} = \langle 0, 0, v_0\rangle$, where $v_0 = 5$ m/s is the design speed. The cross-section $S$ is the disk $x^2 + y^2 \le a^2$ in a plane $z = \text{const}$, oriented downstream with $\hat{\mathbf{n}} = \langle 0,0,1\rangle$.

The flux is immediate, because the integrand $\mathbf{v}\cdot\hat{\mathbf{n}} = v_0$ is constant:

$$Q = \iint_S \mathbf{v}\cdot d\mathbf{S} = \iint_S v_0\,dS = v_0\cdot\text{Area}(S) = 5\cdot\pi(0.20)^2 = 5\cdot 0.04\pi \approx 0.628\ \text{m}^3/\text{s}.$$

This is the elementary "$Q = vA$" formula every fluids textbook prints in chapter one. The surface-integral view reveals why it is so simple: when the field is constant and perpendicular to a flat surface, the flux integral collapses to field strength times area. The whole machinery of $dS$ and $\hat{\mathbf{n}}$ is still there — it has just been trivialized by uniformity. An operating room of volume $150$ m³ would receive $0.628\,\text{m}^3/\text{s}\times 3600\,\text{s/hr} / 150\,\text{m}^3 \approx 15$ air changes per hour from this duct. Not quite the required twenty — so the engineer must either widen the duct or raise the speed, and the flux integral tells her exactly how each choice scales.

The Real Duct: A Velocity Profile

Real air does not move uniformly. Friction at the duct wall enforces the no-slip condition — the air touching the wall is stationary — so the velocity is zero at $r = a$ and maximal on the centerline. For fully developed laminar flow in a round pipe, the profile is the classic Poiseuille parabola:

$$\mathbf{v}(x,y) = \Big\langle 0,\ 0,\ v_{\max}\big(1 - \tfrac{x^2+y^2}{a^2}\big)\Big\rangle.$$

Now the integrand varies across the disk, and the integral genuinely does work. With $\hat{\mathbf{n}} = \langle0,0,1\rangle$, the flux is

$$Q = \iint_S v_{\max}\Big(1 - \frac{x^2+y^2}{a^2}\Big)dA.$$

Switch to polar coordinates, $x^2 + y^2 = r^2$, $dA = r\,dr\,d\theta$:

$$Q = v_{\max}\int_0^{2\pi}\!\!\int_0^a\Big(1 - \frac{r^2}{a^2}\Big)r\,dr\,d\theta = 2\pi v_{\max}\int_0^a\Big(r - \frac{r^3}{a^2}\Big)dr.$$

The inner integral is $\big[\tfrac{r^2}{2} - \tfrac{r^4}{4a^2}\big]_0^a = \tfrac{a^2}{2} - \tfrac{a^2}{4} = \tfrac{a^2}{4}$, so

$$Q = 2\pi v_{\max}\cdot\frac{a^2}{4} = \frac{\pi a^2 v_{\max}}{2} = \frac{v_{\max}}{2}\cdot\text{Area}(S).$$

Here is a result with real engineering bite: the volume flow rate of a parabolic profile equals the centerline speed times half the area. Equivalently, the average velocity across the section is exactly $v_{\max}/2$. An engineer who measures only the (easy-to-reach) centerline speed and multiplies by the full area would overestimate the delivered air by a factor of two — and a ventilation system that delivers half its rated airflow is a clinical hazard. The factor of one-half is not a fudge; it is the honest output of $\iint_S\mathbf{v}\cdot d\mathbf{S}$ for a parabolic field, and it is invisible to the naive $Q=vA$ shortcut.

When the Surface Is Tilted

Sometimes the cross-section the engineer cares about is not perpendicular to the flow. Imagine a flow $\mathbf{v} = \langle 0, 0, v_0\rangle$ passing through a duct branch whose monitoring window is the slanted plane $z = g(x,y) = c - mx$ over a region $D$ in the $xy$-plane. The graph flux formula of Section 36.7, with $\mathbf{F} = \langle P, Q, R\rangle = \langle 0, 0, v_0\rangle$, gives

$$\iint_S \mathbf{v}\cdot d\mathbf{S} = \iint_D\big(-P\,g_x - Q\,g_y + R\big)\,dA = \iint_D v_0\,dA = v_0\cdot\text{Area}(D).$$

Notice what happened: $g_x = -m$ and $g_y = 0$, but because $P = Q = 0$ those slopes never entered. The flux depends only on the shadow area $D$, not on the tilt of the window. Physically, a vertical stream of air carries the same volume per second no matter how you slant the imaginary surface you count it through — tilting the window enlarges its area but reduces the perpendicular component of velocity in exactly compensating proportion. This is the same cancellation that makes the cylinder's vertical-field flux vanish in Section 36.7, and it is a sanity check engineers lean on: a correctly computed flux through a duct is independent of the surface chosen across it, as long as that surface spans the duct. Any two such surfaces give the same $Q$ because the flow between them neither piles up nor disappears — the continuity idea of Section 36.8 made concrete.

From Integral to Instrument

In practice an HVAC technician never writes down $\mathbf{v}(x,y)$. Instead a pitot-tube traverse samples the velocity at a grid of points across the duct, and the flow rate is computed as a discrete sum,

$$Q \approx \sum_{i} v_{z,i}\,\Delta A_i,$$

where $\Delta A_i$ is the area of the patch each probe represents. This is nothing but the Riemann-sum definition of $\iint_S\mathbf{v}\cdot d\mathbf{S}$ — patches, normal components, areas, summed — the same finite-sum approximation that drives the triangulated-mesh flux of Section 36.13. The continuous integral is the exact statement; the traverse is its field-measurable shadow. National standards (ASHRAE, ISO 3966) even prescribe the probe locations as the points that make this Riemann sum most accurate for a turbulent profile. The textbook integral and the certified field procedure are the same mathematics at two resolutions.

Why the Flux View Wins

A student might ask why bother with surface integrals when $Q = vA$ works for the simple case. The answer is that the simple case almost never holds. Real profiles are parabolic or turbulent; real cross-sections are sometimes slanted or non-planar; real ducts branch and merge. The flux integral $\iint_S\mathbf{v}\cdot d\mathbf{S}$ is the one formula that handles all of them, degrading gracefully to $vA$ exactly when uniformity permits. It also connects forward: the statement that flow rate is the same through every cross-section of a sealed duct is the integral form of conservation of mass, which the Divergence Theorem of Chapter 37 will turn into the differential continuity equation $\nabla\cdot(\rho\mathbf{v}) = 0$. The duct calculation is a first, tangible encounter with a principle that governs everything from blood flow to jet engines.

Discussion Questions

In the parabolic-profile calculation, the average velocity came out to exactly $v_{\max}/2$. Why is the average not the simple midpoint of $0$ and $v_{\max}$ along a radius? (Hint: more area lives near the wall than near the center; the polar weight $r\,dr$ matters.)
The tilted-window flux depended only on the shadow area $D$. Construct a field $\mathbf{v}$ for which the tilt would matter, and explain which component of $\mathbf{v}$ makes the difference.
A technician measures only the centerline velocity of a parabolic flow and reports $Q = v_{\max}A$. By what factor is the delivered air over- or under-stated, and why is this dangerous in a hospital?
Two engineers compute the flux of the same flow through two different surfaces spanning the same duct and get different numbers. List three modeling errors that could explain the discrepancy (orientation, leakage between the surfaces, a non-spanning surface).
How is the pitot-tube traverse formula $\sum_i v_{z,i}\,\Delta A_i$ related to the definition of $\iint_S\mathbf{v}\cdot d\mathbf{S}$? What does refining the probe grid correspond to mathematically?

Annotated Further Reading

White, F. M. (2021). Fluid Mechanics (9th ed.). McGraw-Hill. Chapter 6 develops the Poiseuille parabola and the $v_{\text{avg}} = v_{\max}/2$ result rigorously from the Navier–Stokes equations; read it to see where the velocity profile this case study assumes actually comes from.
ASHRAE Handbook — Fundamentals (current edition), chapter on Duct Design. The professional reference that turns $\iint_S\mathbf{v}\cdot d\mathbf{S}$ into air-change-rate requirements and traverse procedures; shows the integral living inside a building code.
Çengel, Y. A., and Cimbala, J. M. (2018). Fluid Mechanics: Fundamentals and Applications (4th ed.). McGraw-Hill. The "volume flow rate as a surface integral" derivation appears early and is unusually explicit about the flux notation, making it a clean bridge from this chapter to engineering practice.
ISO 3966: Measurement of fluid flow in closed conduits — Velocity area method using Pitot static tubes. The standard that prescribes the discrete-sum traverse; read its grid specification as a real-world Riemann-sum rule for surface integrals.