All lessonsLesson 13
How safe is safe? Heyman’s geometric factor
Limit analysis, the safe theorem, and the factor by which an arch could be thinner.
Lessons 1 and 2 drew the line of thrust through an arch by Ungewitter’s construction. An arch could stand as long as some line fitted inside the ring, but the Lehrbuch asks for more: the most favourable line should keep to the middle third of every joint. How much safety does that demand buy, and is it the right demand at all? The modern answer comes from Jacques Heyman, whose paper The Stone Skeleton (1966) recast the old graphic statics of masonry as limit analysis.
Three assumptions
Heyman idealises masonry by three assumptions. It has no tensile strength: a joint can open but cannot pull. It has, for practical purposes, unlimited compressive strength: the stresses in a cathedral are an order of magnitude below crushing. And its joints do not slide. Under these assumptions the theorems of plastic theory, worked out for steel frames in the 1940s and 50s, apply to stone. The most useful of them is the safe theorem: if any line of thrust in equilibrium with the loads can be found that lies wholly within the masonry, the structure will not collapse.
The theorem does not ask which line the arch “really” has. That line depends on how the centring was struck, on the slightest spreading of the abutments, on cracks that nobody sees, and it cannot be known. It is enough that one admissible line exists. Ungewitter reasons differently. For him the line of thrust is a definite path of pressure, and every joint along it should stay closed.
From the Lehrbuch
„Nur wenn der resultierende Druck in dem Kern angreift, bekommt jedes Stück der Fläche eine Druckpressung, soll solches erzielt werden, so darf sich also der Druck … nur im mittleren Drittel bewegen.“Only if the resultant pressure acts within the kern does every part of the joint receive compression; to achieve this, the pressure may move only within the middle third.Ungewitter 1890, p. 142, Figs. 377–378This is an elastic criterion. It comes from the theory of beams and columns of Navier’s time, where a material must not be stressed in tension anywhere. Heyman keeps the drawing but changes its meaning. A joint that opens on one side is not a failure; it is a hinge, and an arch needs four or five hinges before it can move at all.
Every line that fits
A thick arch admits a whole family of lines. Ungewitter draws its two limits, the steepest and the flattest line (p. 130), but keeps both within the kern. Heyman lets them reach the faces of the ring. The steepest line that still fits gives the least thrust: the passive state, which the arch takes up as soon as its abutments spread a little. The flattest gives the greatest thrust: the active state, if the abutments were pushed together. Any thrust in between is possible. Thin the ring in the figure and watch the tinted zone of admissible lines narrow.
GSF 1.69: it stands, with less margin than Heyman asks for. Ungewitter would thicken it: the line leaves the kern. GSF: 1.69; d_min: 0.473 m; H: 42.9–65.1 kN; 1890 factor: 1.69.
- GSF
- 1.69
- d_min
- 0.473 m
- H
- 42.9–65.1 kN
- 1890 factor
- 1.69
At some thickness only one line is left, and it touches the faces of the ring at five points: the extrados at the crown, the intrados on both haunches, the extrados at the two springings. There the joints are about to open into hinges. Any thinner, and no line fits: the five hinges turn the arch into a mechanism and it falls. For the semicircle under its own weight this minimum thickness is 0.1075 times the radius of the centre line, with the haunch hinges about 54.5° from the crown. Milankovitch found the value in 1907; Heyman’s simpler approximation gives 0.106. The figure reproduces it.
A factor on geometry
Heyman measures safety by this margin. The geometric factor of safety is the actual thickness divided by the minimum: GSF = d / dmin. A factor of 2 means the ring could be half as thick and still, just, stand. Heyman suggests about 2 for arches. The factor is a ratio of lengths, so it does not change with size: an arch twice as large in every dimension is exactly as safe. This is why proportional rules of thumb work for masonry (Huerta, “Galileo was wrong”, 2006). It is stability that governs, not strength.
From the Lehrbuch
„Bei Druck durch die Kante erfolgt Umsturz! Bei Druck durch die Kerngrenze ist das Widerlager gesichert, falls die Kantenpressung nicht zu gross ist.“With pressure through the edge, overturning follows! With pressure through the kern limit, the abutment is secure, provided the edge pressure is not too great.Ungewitter 1890, pp. 150–151Read with Heyman’s eyes, the kern rule is a geometric factor too. If the line must stay within the middle third, the joint is three times as deep as the band the line needs: a factor of about 3. Ungewitter therefore demands more than Heyman does. Where the book accepts an arch, the modern check accepts it as well; but many arches that the book would thicken stand with a comfortable margin. The Lab’s modern view shows both verdicts side by side.
The form matters as much as the thickness. The second figure keeps the ring at 0.5 m and changes only its form.
The Semicircle could be as thin as 0.108 × its radius: GSF 1.09 at this thickness. GSF: 1.09; d_min: 0.457 m; d_min : R = 0.1075; H: 30.8–32.9 kN; 1890 factor: 1.1; hinges: extrados 0°, intrados 55°, extrados 90°.
- GSF
- 1.09
- d_min
- 0.457 m
- d_min : R
- 0.1075
- H
- 30.8–32.9 kN
- 1890 factor
- 1.1
- hinges
- extrados 0°, intrados 55°, extrados 90°
The pointed arch lies much nearer the shape of its own line of thrust. The equilateral arch needs only about 0.04 of its radius: a value this tool derives, since the literature gives no standard figure for it. Its hinges also change places: the line touches the intrados at the crown and the extrados on the haunches. The lancet is steeper again but heavier towards the crown, and its margin falls. Ungewitter’s remedy for the slender lancet, a load at the crown (Lesson 2), is exactly what moves its line back towards the middle.
Challenge
In the Lab’s modern view, thin the semicircular ring until the geometric factor is 2. Then switch to 1890: does Ungewitter accept that ring? Now find the ring he would accept, and read off its geometric factor.