LOAD PATH
The Voussoir Arch
stands down to t = 0.1075 R
A ring of wedge-shaped stones stands as long as a line of thrust fits inside it; Jacques Heyman made that the modern safety check. A semicircular arch needs a ring only 0.1075 of its radius deep, the thrust grazing the intrados 54.5° from the crown. This one is nearly twice that; its keystone is simply the last stone set.
LOAD PATH
Gaudí’s Hanging Model
Colònia Güell, scale 1:10
Gaudí designed the Colònia Güell church upside down, refining a one-tenth-scale model of strings and sachets of lead shot for about ten years from 1898. A string can only pull, so each settles into the shape that carries its loads in tension; inverted, every member is in pure compression. Only the crypt was built.
LOAD PATH
Euler Buckling and Effective Length
K = 2, 1, 0.7, 0.5
Euler found in 1744 the load at which a slender column bows sideways. A flagpole buckles as a quarter wave at a quarter of the pinned-end load; clamped ends take four times it. Fixed and pinned, tan kL = kL gives K = 0.699. Codes raise the clamped values to 0.65 and 0.80, since no connection is perfectly rigid.
LOAD PATH
Parabola, Not Catenary
hangers grow as x²
A free chain hangs as a catenary, its weight spread along its length. Hang a level deck from it and the load is spread along the span instead: the curve becomes a parabola, and the hangers grow as the square of their distance from the center. Every suspension cable is designed that way.
LOAD PATH
Warren and Pratt Trusses
Pratt diagonals: all tension
The Pratts patented their truss in 1844, Warren and Monzani theirs in 1848. The method of joints solves either, two equations at every joint. Loaded at its inner joints, the Pratt puts every diagonal in tension, 3.54, 2.12 and 0.71 times the joint load, and its center post and end bottom panels carry nothing: zero-force members.
MAIN SPAN
Brooklyn Bridge Cable Spinning
19 strands of parallel wire
Brooklyn’s four main cables, each 15¾ inches across, were spun in the air in 1877 and 1878. A wheel on an endless rope carried a loop of wire across on every trip, each wire set to the sag of a guide wire. They were bundled into 19 strands, one, six and twelve, then squeezed round and wrapped.
MAIN SPAN
Cable-Stayed: Fan and Harp
harp: 8.6 % more stay steel
A cable-stayed deck hangs straight from its pylon, each stay carrying its share with a tension of ws/sin θ, so the flatter the stay, the harder it works. A fan runs every stay to the pylon head and keeps them steep. A harp keeps them parallel, which looks calm, but here needs 8.6% more stay steel.
MAIN SPAN
Bridge Scour
up to 2.4 pier widths deep
When a river meets a pier it plunges down the upstream face and rolls into a horseshoe vortex that lifts sand faster than the flow returns it. The Federal Highway Administration’s HEC-18 equation predicts the hole: 1.93 pier widths for a round pier in water three widths deep at a Froude number of 0.3, capped at 2.4.
MAIN SPAN
Tacoma Narrows, 1940
one node at midspan, 0.2 Hz
Tacoma Narrows opened on 1 July 1940: a 2,800-foot span, a deck only 39 feet wide on plate girders 8 feet deep. On 7 November, in a wind of about 42 mph, it twisted in its first antisymmetric torsional mode until the span tore loose. Not resonance: flutter, energy drawn from a steady wind.
MAIN SPAN
The Forth Bridge Cantilever
arms pull, sticks push
Baker explained the Forth Bridge with a photograph in 1887. Two men sit with arms out, gripping sticks braced against their chairs: arms in tension, sticks in compression, chairs as piers. Kaichi Watanabe, one of the engineers, sits between them as the suspended span, and piles of bricks hold down the outer arms.
28 DAYS
The Cylinder Break
f ′c = 4P/πd² at 28 days
Standard cylinders, 150 by 300 mm or 100 by 200, are cast from the truck, cured and broken at 28 days; at 30 MPa a 150 mm cylinder fails at 530 kN. Platen friction confines the ends, so a good break is two cones. A diagonal shear or a split through a cap tells the technician something went wrong.
28 DAYS
The Pantheon Dome
43.3 m span, 5 × 28 coffers
Hadrian’s Pantheon, finished around AD 125, would hold a sphere 43.3 m across. Five rings of 28 coffers lighten the shell, and the aggregate grades from heavy stone to light pumice near the top. Under its own weight a hemisphere’s hoops go into tension beyond 51.8° from the crown, and the lower dome is cracked in just that zone.
28 DAYS
The Slump Test
300 mm cone, 75 mm slump
A steel cone 300 mm tall, 200 mm across the base and 100 across the top, filled in three layers rodded 25 times each and lifted up. The drop is the slump, 75 mm here. A true slump keeps its form; a shear slump slides and a collapse spreads, and each says something different about the mix.
28 DAYS
Confinement and the Seismic Hook
135° hooks into the core
In an earthquake a column must hold its strength while bending far past yield. Closed hoops confine the core and stop the bars buckling once the cover spalls, but only if they stay shut: a 90° hook opens, while a 135° hook bent into the core with a tail of at least six bar diameters or 75 mm holds.
28 DAYS
Post-Tensioning and Load Balancing
upward load w = 8Pe/L²
Freyssinet’s answer to concrete’s weakness in tension, patented in 1928, was to squeeze it first. Drape the tendon in a parabola and, pulled tight, it pushes up with an even load, 8Pe/L². T. Y. Lin’s 1963 method cancels the self-weight with it: a 20 m beam with a 2,000 kN tendon draped 300 mm gets 12 kN per meter.
REFUSAL
Mohr–Coulomb and the Triaxial Test
fails on a plane at 45° + φ′/2
In the triaxial test a sealed soil cylinder, twice as tall as wide, is squeezed by water in a clear cell and loaded by a ram. A sand with a 30° friction angle fails at three times the cell pressure, on a plane at 45° plus half that angle: 60°. Bishop and Henkel’s 1957 book made it standard.
REFUSAL
The Leaning Tower of Pisa
eased back ½° in 1999–2001
Work stopped in 1178 with three stories done, and the pause probably saved the tower, letting the clay consolidate. Builders resuming in 1272 made the upper stories taller on the low side, hence the curve. Jamiolkowski’s commission, with Burland on the soil, eased it back about half a degree by 2001, leaving a lean of about 4°.
REFUSAL
The Pressure Bulb
0.1q reaches 2B deep
Boussinesq solved the stress under a point load on an elastic half-space in 1885. Summed under a square footing of width B, the stress falls to a half, a fifth and a tenth of the footing pressure at depths of 0.73B, 1.40B and 2.09B. That last depth is the borehole rule: investigate about two widths down.
REFUSAL
The Liquid Limit
13 mm closes at 25 blows
Atterberg defined the limits of a clay in 1911; Casagrande made the liquid limit a machine. A cam lifts the brass cup 10 mm and drops it twice a second; the liquid limit is the water content at which 25 blows close the groove along 13 mm. By the one-point method, 42% water at 30 blows means 42.9%.
REFUSAL
Seepage into a Cofferdam
4 flow channels, 10 drops
Groundwater obeys Laplace’s equation, so seepage under a sheet-pile wall can be drawn as a flow net of curvilinear squares, as Forchheimer did in 1911. Leakage is permeability times head times channels over drops. Here: four channels, ten drops a side. The exit gradient beside the pile is 0.43; near 1 the floor boils.
FREEBOARD
The Morning-Glory Spillway
radial inflow, no vortex
A morning-glory spillway drains a full reservoir through a funnel in the lake: a circular crest, a bell-mouth, a vertical shaft and a tunnel. It suits narrow valleys and works as a weir bent into a circle. The best known is the Glory Hole at Monticello Dam in California, 72 feet across, completed in 1957.
FREEBOARD
The Egg-Shaped Sewer
height 1½ × width
A sewer must carry storms yet keep a dry-weather trickle moving fast enough to carry its solids. The Victorian egg stands on its narrow end, crown radius half the width, invert a quarter. At 2% of full flow, Manning’s equation puts the water 0.154 of the width deep, moving 10.5% faster than in a round pipe of equal area.
FREEBOARD
The Parshall Flume
one depth reading gives Q
Ralph Parshall’s 1920s flume measures open-channel flow from one depth. The throat floor drops steeply, so the flow passes critical depth and the upstream depth depends on the flow alone. A one-foot throat passes four cubic feet per second at a depth of one foot, 11.5 at two, and silt that would clog a weir.
FREEBOARD
The Hydraulic Jump
Fr 5: the depth jumps 6.6×
When fast, shallow water meets slow, deep water it jumps. Bélanger’s momentum balance makes the depth ratio depend only on the upstream Froude number: at 5 the water rises 6.59 times, through a churning roller, and loses 49% of its energy. Stilling basins hold a jump in place so that energy is spent on concrete, not on the riverbed.
FREEBOARD
The Miter Gate
the V always points upstream
Two leaves swing from the lock walls and meet in a shallow V pointing upstream, so the higher water pushes them together and the gate seals tighter the more it holds back. Each leaf is a strut carrying a thrust of pW/4 sin θ into the quoin. A rise of one in three, 18° off square, is the traditional compromise.