Case Study 2 — Designing a Curve You Can Take at Speed

Field: Civil and mechanical engineering (highway and roller-coaster design) Calculus used: Curvature $\kappa$ (§28.8), the normal component of acceleration $a_N = \kappa v^2$ (§28.9), radius of curvature and the osculating circle (§28.8), and the clothoid as a curve of linearly varying curvature (§28.9).

The feeling in your stomach is a vector

Priya Desai designs alignments for a state highway authority, and on weekends she consults for an amusement park redesigning a roller coaster. The two jobs look unrelated — a gentle interstate sweep versus a loop that flips riders upside down — but to Priya they are the same equation wearing different clothes. The sideways shove a driver feels rounding a bend, and the crush a rider feels at the bottom of a loop, are both the normal component of acceleration from §28.9:

$$a_N = \kappa\,v^2,$$

the product of how sharply the path bends ($\kappa$) and the square of how fast you take it ($v$). Everything in this case study is a consequence of that single formula and of the curvature machinery the chapter built around it. The chapter's headlight image makes it concrete: at every instant your path hugs an osculating circle of radius $\rho = 1/\kappa$, and your body is being pulled toward its center.

Why a fast road must be a gentle road

Priya's first task is a new freeway interchange rated for 30 m/s (about 108 km/h, 67 mph). Human comfort and tire grip cap the lateral acceleration; a common design target for a high-speed road is about $0.12\,g$, where $g = 9.8\ \text{m/s}^2$. So the requirement is

$$a_N = \kappa v^2 \le 0.12\,g \approx 1.18\ \text{m/s}^2.$$

Because $a_N$ scales with the square of speed, this is a harsh constraint at freeway speeds. Solving for the curvature ceiling at $v = 30$ m/s:

$$\kappa \le \frac{0.12\,g}{v^2} = \frac{1.18}{900} \approx 1.31\times 10^{-3}\ \text{m}^{-1}.$$

In radius-of-curvature terms (§28.8), $\rho = 1/\kappa$, so the minimum radius is

$$\rho_{\min} = \frac{v^2}{0.12\,g} = \frac{900}{1.18} \approx 765\ \text{m}.$$

A freeway bend must be at least a three-quarter-kilometre arc — the long, lazy sweeps you see on interstates. Now contrast a 12 m/s (27 mph) city street: the radius requirement collapses to $\rho_{\min} = 144/1.18 \approx 122$ m, six times tighter, because curvature scales with $1/v^2$. This is precisely the §28.9 point: a road rated for high speed needs very large radii. Priya does not memorize a table; she reads it off $a_N = \kappa v^2$.

A quick sanity check from the chapter's exercises: a car taking a flat $R = 50$ m bend at 20 m/s feels $a_N = v^2/R = 400/50 = 8\ \text{m/s}^2 \approx 0.82\,g$ — far past the comfort target. That bend is fine for a slow ramp but would throw coffee across the cabin at highway speed. The radius and the speed must be designed together.

Banking: tilting the road to help

A flat curve forces friction to supply the entire sideways force. Priya can do better by banking the roadway — tilting it inward so that a component of gravity does the turning. On a frictionless banked curve, the geometry balances when

$$\tan\theta = \frac{v^2}{gR} = \frac{a_N}{g}.$$

For the freeway curve at the limiting radius $R = 765$ m and $v = 30$ m/s,

$$\tan\theta = \frac{900}{9.8 \times 765} = \frac{900}{7497} \approx 0.120, \qquad \theta \approx 6.8^\circ.$$

A bank of roughly seven degrees lets a car at design speed round the curve with almost no reliance on tire friction — leaving the friction budget as a safety margin for rain, ice, or a driver going a little fast. Notice that $\tan\theta = a_N/g$ is just the normal acceleration re-expressed as an angle: the steeper the required turn, the harder the road must lean. The same principle, exaggerated, is why a banked velodrome or a NASCAR oval tilts so dramatically.

The clothoid: curvature you can't switch on instantly

Here is the subtlety that separates a comfortable road from a jarring one, and it is pure §28.9. Suppose Priya joined a straight tangent section ($\kappa = 0$) directly to a circular arc ($\kappa = 1/765$). At the join, curvature would jump discontinuously from $0$ to its full value. Since $a_N = \kappa v^2$, the lateral acceleration would jump too — from nothing to $1.18\ \text{m/s}^2$ in an instant. The driver experiences that jump as a sharp sideways jerk, and must yank the steering wheel to a fixed angle the moment the arc begins.

Real highway and railway curves therefore insert a transition spiral, almost always a clothoid (Euler spiral), whose defining property is exactly what §28.9 names: its curvature increases linearly with arc length,

$$\kappa(s) = \frac{s}{A^2},$$

ramping smoothly from $0$ at the start of the spiral up to the circular arc's value. Because $a_N = \kappa v^2$ and (at roughly constant speed) $s \approx vt$, the lateral acceleration now rises linearly in time rather than jumping. The rate of change of that acceleration is the jerk — and keeping jerk finite and small is the whole point. In single-variable language, jerk is the third derivative of position, $\mathbf{r}'''$; the clothoid is engineered so that $da_N/dt$ stays bounded, letting the driver turn the wheel at a steady rate instead of all at once. The same spiral lets a railway lift the outer rail gradually so the train rolls smoothly into the bank.

The roller coaster: the same formula, dialed up

On the weekend job, Priya turns the formula in the opposite direction. A vertical loop is designed to deliver a large $a_N$ on purpose — that pressed-into-the-seat feeling is the product. Take a circular loop of radius $R = 10$ m, with the train passing the bottom at $v = 15$ m/s. Using $\kappa = 1/R$,

$$a_N = \kappa v^2 = \frac{v^2}{R} = \frac{225}{10} = 22.5\ \text{m/s}^2 \approx 2.3\,g.$$

But at the bottom of the loop the rider also carries their own weight, so the seat must push up with the centripetal force plus gravity: the felt load is roughly $2.3\,g + 1\,g \approx 3.3\,g$. That is near the comfortable ceiling for a public ride; design limits usually live around $4$–$5\,g$ and only briefly. Push the speed up and, because $a_N \propto v^2$, the g-load climbs fast — which is exactly why modern coasters almost never use a circular loop. They use a clothoid loop: tall and pinched at the top, with the radius (and thus $1/\kappa$) varying around the loop so that the rider meets large curvature only where the speed is low (near the top), and small curvature where the speed is high (near the bottom). The product $\kappa v^2$ is held roughly constant all the way around, smoothing the g-force into a steady plateau instead of a violent spike. The very same curve that makes a highway gentle makes a loop survivable — both are applications of keeping $\kappa v^2$ under control.

One equation, two trades

Priya's whole craft reduces to managing $a_N = \kappa v^2$ and its time rate of change. On the highway she minimizes $a_N$: big radii, modest banks, clothoid transitions so the curvature — and the jerk — ramp up gently. On the coaster she maximizes $a_N$ within human tolerance, then shapes the curve so it never spikes. In both cases the design variables are the two factors the chapter isolated: the curvature of the path (a property of the geometry alone, §28.8) and the speed at which it is taken. The osculating circle of §28.8 is not an abstraction to her; it is the turning radius her ruler draws, and the force her riders feel is its inverse, squared against the speed.

Discussion Questions

1. The freeway needed a 765 m radius but a 12 m/s city street needed only about 122 m. Show, from $\rho_{\min} = v^2/a_{N,\max}$, why halving the design speed quarters the required radius. Why does this make high-speed rail and freeways look so different from city streets on a map?
2. The banking relation $\tan\theta = a_N/g$ assumed no friction. Qualitatively, how does available tire friction change the range of speeds a fixed banked curve can safely serve — and why is a single "design speed" a compromise?
3. A clothoid has $\kappa(s) = s/A^2$. Explain, using $a_N = \kappa v^2$ at constant speed, why this makes the lateral acceleration grow linearly in time and the jerk roughly constant. What would happen to the jerk at a straight-to-circle join with no transition?
4. For the circular coaster loop, the felt load at the bottom was $a_N + g$. What is the felt load at the top of the loop, and why must the speed there stay above a minimum for the train to maintain contact with the track?
5. Both projects manage the same quantity $\kappa v^2$, one minimizing it and one maximizing it. Identify which chapter formula supplies $\kappa$ for a circle, and which supplies it for a general space curve (§28.8) — and why the coaster designer needs the general one.

A Short Annotated Reading

• AASHTO (2018). A Policy on Geometric Design of Highways and Streets ("the Green Book"). The engineering bible for horizontal curves, superelevation (banking), and spiral transitions; its design tables are $a_N = \kappa v^2$ in disguise.
• Stewart, J. Calculus: Early Transcendentals, §13.3 "Arc Length and Curvature." The textbook treatment of $\kappa$, the osculating circle, and the tangential/normal split that underlies every number in this case study.
• Levin, A. (2002). "The mathematics of roller coasters," The Physics Teacher. A readable account of why real loops are clothoids rather than circles, framed exactly around holding $\kappa v^2$ steady.