CEE 330 · Structural Analysis and Design-I · Week 1

Arches & Cables

A 40 m beam under its design load carries 4000 kN·m of bending. Bend that same beam into a parabola, hold its feet from spreading, and the bending drops to zero. Not "small" — zero. This chapter is about why.

Funicular shape Three-hinged arch Horizontal thrust Cable sag & tension
01

Why Bother Curving a Beam?

A beam in bending is a structure at war with itself. The top fibres squash, the bottom fibres stretch, and the middle does almost nothing — you are paying for material that is barely working. An arch is the attempt to make every fibre carry the same compression, all the way through.

🥋
Carrying a stack of books. Hold the stack out in front of you, arms horizontal, and your muscles scream — that is bending, and the load is fighting your arm sideways. Now hold the stack directly overhead with straight arms. You can stand there for minutes: the force runs down the bone, and bone in pure compression is enormously strong. An arch does exactly this. It arranges its own geometry so the load always runs down the material's length, never across it.
This is why masonry arches exist and masonry beams do not. Stone, brick and concrete are strong in compression and feeble in tension. A stone beam cracks on its underside almost immediately. A stone arch, shaped correctly, never puts a single fibre in tension — which is how bridges built before anyone could calculate a stress are still carrying traffic today.
The price

Arches push outwards

A beam presses straight down on its supports. An arch also shoves them apart, with a horizontal force called the thrust. Every gain in this chapter is bought with that thrust, and if the abutments cannot resist it, the arch flattens and fails.

The prize

Bending goes away

Get the shape right for the load and the bending moment vanishes over the whole span. The section can then be slim, because it only has to resist axial compression — no lever arm required.

02

The Hanging Chain

Before computing anything, there is one physical fact that gives you the entire chapter for free.

The key observation

A chain cannot resist bending — it has no stiffness at all. So when you hang it between two points, it has no choice but to settle into the one shape in which the load is carried by pure tension. That shape is not a guess. It is the only shape that works.

🔗
Hooke's anagram, 1675. Robert Hooke published his discovery as a scrambled Latin anagram so he could claim priority without revealing it. Decoded: "Ut pendet continuum flexile, sic stabit contiguum rigidum inversum""as hangs the flexible line, so but inverted will stand the rigid arch." Hang a chain, photograph it, turn the photo upside down, and build that. Every tension in the chain becomes a compression in the arch, of exactly the same magnitude. Gaudí designed the Sagrada Família this way, with a ceiling of hanging strings and weights and a mirror on the floor.
Load

Uniform along the span

A deck of constant weight per horizontal metre. The chain hangs as a parabola. This is the case that dominates suspension bridges and this course.

Load

Uniform along the cable

The cable's own weight, with nothing hanging from it. The shape is a catenary (a cosh curve). For shallow sags it is within a percent of the parabola, so we use the parabola.

Load

Point loads

The chain becomes a chain of straight segments with a kink at each load. Flip it and you get the polygonal arch — still no bending, just a different funicular polygon.

The word to remember is funicular (from Latin funiculus, "thin rope"). The funicular shape is the shape that carries a particular load with no bending. Change the load and the funicular shape changes — but the arch cannot change with it. That single mismatch is the source of every bending moment in Section 7.
03

Cables: Sag, Thrust and Tension

Take a cable of span L carrying a uniform load w per horizontal metre, sagging a distance d at midspan. Cut it at the lowest point and look at half of it.

1
At the low point the cable is horizontal, so the force it transmits there is purely horizontal. Call it H. It is the same H everywhere along the cable, because nothing horizontal is applied in between.
2
Take moments about the support for the half-cable: the load on that half is \(wL/2\) acting at \(L/4\) from the low point, and H acts with lever arm d.
\[ H\,d = \frac{wL}{2}\cdot\frac{L}{4} \quad\Rightarrow\quad H = \frac{wL^2}{8d} \]
3
Vertical component at the support is just half the load: \( V = wL/2 \).
4
Tension is the resultant, and it is largest at the supports where the cable is steepest:
\[ T_{\max} = \sqrt{H^2 + V^2}, \qquad T_{\min} = H \ \text{(at the low point)} \]
Read the formula for H and a design rule falls out. H is inversely proportional to sag. Halve the sag and you double the cable force and the pull on the towers. A taut cable looks efficient and is the opposite — that is why suspension bridges sag as deeply as clearance allows, typically around L/10.
Cable length

The developed length of a shallow parabolic cable, which you need to order the material and to compute stretch:

\[ S \approx L\left(1 + \frac{8}{3}\left(\frac{d}{L}\right)^2 - \frac{32}{5}\left(\frac{d}{L}\right)^4\right) \]

For d/L = 0.1 this gives S = 1.0260 L — the cable is only 2.6% longer than the straight line. Sag is cheap in material and enormously valuable in force.

Drag the sag-ratio slider and watch Tmax: at d/L = 1/2 the cable is nearly slack and the force is small; at d/L = 1/30 it is almost straight and the force explodes. The load never changed.

04

Flip It: The Three-Hinged Arch

Now invert the cable. Tensions become compressions, the downward sag becomes an upward rise h, and the horizontal pull H becomes the horizontal thrust pushing outwards on the abutments.

Why three hinges

To make it determinate

Two pinned supports give four reaction components (VA, HA, VB, HB) against three equations — one redundant. Put a hinge at the crown and you gain one equation of condition (\(\Sigma M = 0\) about the hinge, for one side only). Four unknowns, four equations: determinate, and solvable with statics alone.

Why it matters practically

Immune to settlement

A two-hinged or fixed arch develops large forces if an abutment settles or the temperature changes, because it is indeterminate. The three-hinged arch simply adjusts its shape and carries on. That robustness is why it is the standard starting point.

The whole calculation, in two moves

Move 1 — vertical reactions: take moments about A and B for the whole arch. These come out exactly as for a simply supported beam of the same span, because H acts through both supports at deck level and contributes nothing.

Move 2 — the thrust: take moments about the crown hinge C, using only the left half. The internal moment there is zero by definition of a hinge:

\[ H = \frac{M_{C,\text{beam}}}{h} \]

where \(M_{C,\text{beam}}\) is the bending moment that a simply supported beam of the same span would have at midspan under the same load. For a full UDL that is \(wL^2/8\), giving the formula you will use constantly:

\[ \boxed{\,H = \frac{wL^2}{8h}\,} \]
The thrust is a lever bargain. Something has to supply the moment \(M_{C,\text{beam}}\). A beam supplies it internally, with a couple across its own depth. An arch supplies it externally, with the horizontal thrust H acting on the lever arm h. A tall arch has a long lever and needs little thrust; a flat arch has a short lever and needs an enormous one. The same trade you already saw with cable sag — because it is the same equation upside down.
And here is the punchline. The bending moment at any section of the arch is \( M = M_{\text{beam}}(x) - H\,y(x) \). If the arch axis y(x) is shaped so that \(H\,y(x)\) matches the beam moment diagram at every point, then M = 0 everywhere. For a UDL the beam moment diagram is a parabola — so the funicular arch is a parabola. Not a coincidence: the funicular shape is the beam's own moment diagram, scaled by 1/H.
05

Simulator: Arch vs Beam

The top panel is the structure with its reactions; the bottom panel is the bending moment along the span, drawn to the same scale for all three shapes so you can see them collapse. Start with the parabolic arch under UDL only, then switch the shape to "straight beam" and watch the moment diagram appear from nothing.

Try this

Parabola, UDL only

Set P = 0. The moment diagram is a flat line at zero for the whole span. The arch is carrying 800 kN of load with no bending at all.

Try this

Switch to circular

Same span, same rise, same H — but the axis no longer matches the parabola, so a moment appears. It peaks near the quarter points, at about 4% of the beam value.

Try this

Add a point load

Set P = 100 kN at a = 10 m. A kink appears under the load. The UDL is still carried funicularly — only the mismatched part of the load causes bending.

06

Forces at a Section

An arch section carries three things. Cut the arch at horizontal position x, keep the left part, and resolve everything that acts on it into components along and across the arch axis — not vertical and horizontal, because the axis is sloped.

\[ \tan\theta = \frac{dy}{dx} = \frac{4h(L-2x)}{L^2} \quad\text{(parabolic axis)} \]
Normal thrust

N — along the axis

\[ N = H\cos\theta + V\sin\theta \]

Compression. This is the force the section actually lives on. Largest at the springings, where the axis is steepest.

Radial shear

Q — across the axis

\[ Q = V\cos\theta - H\sin\theta \]

Zero everywhere for a true funicular arch — the resultant lies exactly along the axis, so it has no across-component.

Bending moment

M — the mismatch

\[ M = M_{\text{beam}}(x) - H\,y(x) \]

V here is the net vertical force to the left of the cut — the same "beam shear" you already know how to compute.

🏹
The thrust line. At every section the resultant of H and V passes through some point on the cross-section. Join those points along the arch and you get the thrust line — where the load is really flowing. If the thrust line lies exactly on the arch axis, M = 0. If it drifts off the axis by a distance e, then M = N·e. So the bending moment is nothing more than the eccentricity of the thrust line from the axis, and a masonry arch is safe as long as that line stays inside the stonework.
Check yourself with the simulator. Set the parabola, UDL only, and drag the section slider to the springing (x = 0). You should read N = 640.3 kN, Q = 0, M = 0 — and 640.3 is exactly \(\sqrt{400^2+500^2}\), the resultant of the two reactions. The whole reaction is running straight into the arch.
07

When an Arch Does Bend

Real arches always bend a little, and there are only three reasons why.

Reason 1

The load changed

The arch was shaped for its dead load. Traffic on half the span, snow drifting to one side, or a single heavy vehicle is a different load with a different funicular shape. Asymmetric live load is usually the governing case.

Reason 2

The shape is wrong

Circular arches are easier to build than parabolic ones, so they get built. The 4% residual moment in the simulator is the price of that convenience — usually acceptable, and always worth knowing about.

Reason 3

The thrust was not delivered

If the abutments spread, H drops, and the term \(H\,y\) can no longer cancel the beam moment. The arch starts behaving like a curved beam — and curved beams are bad beams. This is how old masonry bridges fail.

Core Idea

Everything in this chapter is one equation seen from different angles:

\[ M(x) = \underbrace{M_{\text{beam}}(x)}_{\text{what the load demands}} - \underbrace{H\,y(x)}_{\text{what the geometry supplies}} \]

Match them and the arch is pure compression. Mismatch them and the difference shows up as bending. The funicular shape, the thrust formula, the sag–force trade and the thrust line are all just readings of this one balance.

08

Worked Examples

Example 1

Three-Hinged Parabolic Arch under UDL

Problem: A three-hinged parabolic arch spans L = 40 m with a crown rise h = 8 m and carries a UDL of w = 20 kN/m over the full span. Find the reactions, the maximum bending moment, and the thrust at the springing.

1
Vertical reactions — symmetric, so \( V_A = V_B = \dfrac{wL}{2} = \dfrac{20 \times 40}{2} = 400\text{kN} \)
2
Horizontal thrust from the crown hinge: \( H = \dfrac{wL^2}{8h} = \dfrac{20 \times 40^2}{8 \times 8} = \dfrac{32000}{64} = 500\text{kN} \)
3
Bending moment at any x. The parabolic axis is \( y = \dfrac{4hx(L-x)}{L^2} \) and the beam moment is \( \dfrac{wx(L-x)}{2} \):
\[ M = \frac{wx(L-x)}{2} - \frac{wL^2}{8h}\cdot\frac{4hx(L-x)}{L^2} = \frac{wx(L-x)}{2} - \frac{wx(L-x)}{2} = 0 \]
4
Springing angle: \( \tan\theta = \dfrac{4h}{L} = \dfrac{32}{40} = 0.8 \Rightarrow \theta = 38.66^\circ \)
5
Normal thrust there: \( N = H\cos\theta + V\sin\theta = 500(0.7809) + 400(0.6247) = 390.4 + 249.9 = 640.3\text{kN} \), and \( Q = 400(0.7809) - 500(0.6247) = 0 \).
Answer: V = 400 kN, H = 500 kN, M = 0 everywhere, N = 640.3 kN at the springing with zero shear. A simply supported beam of the same span would carry \(wL^2/8 = 4000\) kN·m of bending. The arch replaces all of it with 500 kN of horizontal push.
Example 2

The Same Arch under a Point Load

Problem: Remove the UDL and apply a single P = 100 kN at a = 10 m from A. Find H and the bending moment under the load.

1
Vertical reactions as for a beam: \( V_A = \dfrac{P(L-a)}{L} = \dfrac{100(30)}{40} = 75\text{kN} \), \( V_B = 25\text{kN} \)
2
Thrust — moments about the crown, left half only. The load lies on the left half, so it appears:
\[ H h = V_A\frac{L}{2} - P\left(\frac{L}{2}-a\right) = 75(20) - 100(10) = 500 \Rightarrow H = \frac{500}{8} = 62.5\text{kN} \]
3
Arch height under the load: \( y(10) = \dfrac{4(8)(10)(30)}{40^2} = \dfrac{9600}{1600} = 6\text{m} \)
4
Moment there: \( M = V_A x - Hy = 75(10) - 62.5(6) = 750 - 375 = 375\text{kN·m} \)
Answer: H = 62.5 kN and M = 375 kN·m under the load. The equivalent beam would carry \(Pab/L = 750\) kN·m — so the arch halves it, but does not eliminate it. A point load is not the funicular load for a parabola, and the arch tells you so.
Example 3

Cable: Sag, Tension and Length

Problem: A cable spans L = 30 m with a midspan sag of 3 m and carries w = 5 kN/m along the horizontal. Find H, the maximum and minimum tension, and the cable length.

1
\( H = \dfrac{wL^2}{8d} = \dfrac{5 \times 900}{8 \times 3} = \dfrac{4500}{24} = 187.5\text{kN} \) — this is also \(T_{\min}\), at the lowest point.
2
\( V = \dfrac{wL}{2} = \dfrac{5 \times 30}{2} = 75\text{kN} \) at each support.
3
\( T_{\max} = \sqrt{187.5^2 + 75^2} = \sqrt{35156 + 5625} = 201.9\text{kN} \)
4
\( S \approx 30\left(1 + \tfrac{8}{3}(0.1)^2 - \tfrac{32}{5}(0.1)^4\right) = 30(1.02603) = 30.78\text{m} \)
5
Now halve the sag to 1.5 m: \( H = 375 \) kN and \( T_{\max} = 382.4 \) kN — nearly double, for a cable that is only 0.58 m shorter (30.20 m instead of 30.78 m).
Answer: H = Tmin = 187.5 kN, Tmax = 201.9 kN, S = 30.78 m. Note how little T varies along the cable (7.7% here): for shallow cables the tension is nearly uniform, which is why a single cable size works over the whole span.
09

Quick Reference & Quick Check

Funicular shapeshape carrying a given load with zero bending
UDL along spanparabola
Self-weight onlycatenary (≈ parabola if shallow)
Point loadsfunicular polygon (straight segments)
Cable thrustH = wL²/(8d)
Cable Tmax√(H² + V²), at the supports
Cable TminH, at the lowest point
Cable lengthL[1 + (8/3)(d/L)² − (32/5)(d/L)⁴]
Arch thrustH = MC,beam/h; = wL²/(8h) for full UDL
Arch vertical reactionssame as the equivalent simple beam
Arch momentM = Mbeam(x) − H·y(x)
Normal thrustN = H cosθ + V sinθ
Radial shearQ = V cosθ − H sinθ
Thrust lineM = N·e, e = offset from the axis
Three hinges4 unknowns, 4 equations → determinate

1. A three-hinged parabolic arch carries a full-span UDL. What is the bending moment at the quarter point?

2. You halve the sag of a cable while keeping the span and load the same. The maximum tension:

3. Why is a hinge placed at the crown of an arch?

4. The abutments of a masonry arch spread slightly outwards over time. What happens?

5. Hooke's principle "as hangs the flexible line, so but inverted will stand the rigid arch" means: