Skip to content
Ungewitter

RuleModern statics

The arch on spreading supports

modern.kinematics.spreading

documentedimplementedModern

Stated in a medieval text.

Statement

When the supports of an arch move apart, it at once takes its state of least thrust and forms three hinges: one at the extrados of the crown and one at the intrados of each haunch, where the thrust of the crown segment is greatest (Coulomb’s condition). The arch then deforms as a mechanism of rigid blocks: the crown segments turn about their haunch hinges, the lower parts move with the supports, the crown drops and the thrust rises, while the vertical reaction stays half the weight. As the supports spread, the haunch hinges may creep toward the crown, which raises the thrust further. The arch collapses when its thrust is more than the lower parts can carry, i.e. when the line of pressure points reaches the extrados at the springing: five hinges in a symmetrical arch, four with the slightest asymmetry. A very flat, thick arch can instead snap through as a three-hinge mechanism. The span increase an arch survives grows almost linearly with its thickness: none at the least thickness (t/R = 0.1075 for the semicircle), about 8 % of the span for a 120° arch with t/R = 0.10, about 21 % for a semicircle with t/R = 0.24.

Formula

Hmin⁡=maxβ⁡WAC(xA−x‾AC)yC−yA,collapse when H>WAC(xA−xD)+WDA(x‾DA−xD)yA−yD

Parameters

table_5_3.half_angle_deg
60
table_5_3.t_over_R
0.1
table_5_3.voussoir_deg
0.1, 1, 5, 10
table_5_3.span_increase_pct
8.2, 8, 8.6, 9.3
table_5_3.thrust_ratio
2.16, 2.17, 2.02, 2.03
table_5_3.hinge_at_collapse_deg
42.7, 42, 40, 40
table_5_3.source_page
91
reproduction
With ArchSpread’s own stepping (one hinge move per step of 1/1000): Table 5.3 within 0.1 points of span, 1° of hinge and 0.05 t of dip, for both versions below. Fig. 5.15 hinges within 0.3°. The semicircle’s curve of Fig. 5.14 within ±0.3 points of span at 18 digitised points (largest difference 0.19). Table 5.4: Arch1 16.9 % reproduced (dip 1.05 t against the printed 1.01 t).
thrust_ratio
The thesis’s “thrust increase” is the listing’s thrustmax/thrustmin: the capacity Hmax of the supporting segment at the final hinge, with the thrust taken at the intrados hinge A. After a hinge jump it lies below the thrust the crown segment then demands. The engine reports the same quantity (until wave 6 it referred the thrust to the crown hinge, which gives the same collapse but a value between demand and capacity: 2.41 and 2.84 for 5° and 10°). Corrected statics: 2.164, 2.116, 2.020, 2.027 for 0.1°, 1°, 5°, 10°; the printed listing: 2.162, 2.158, 2.223, 2.434. The table (2.16, 2.17, 2.02, 2.03) follows the corrected statics for 5° and 10° to the printed digit, the listing for 1° (within 0.012); no single version gives all four.
listing_quirk
In the printed listing (p. 157) the hinge-move block computes Hmax = (M2(XD − XM2) + M1(XD − XA))/(YA − YD) before it updates M1, so after each move the capacity counts the voussoir the hinge has just passed twice; Hmax is not recomputed until the next move. A literal port (spreadingArch with listing: true) reproduces Table 5.4’s Arch2 prediction: 8.84 % (printed 8.8 %), dip 1.24 t (printed 1.21 t), the capacity at 50° then 1.77 H_min instead of 1.53. With the correct weight the thrust after the jump from 60° to 50° (1.55 H_min) already exceeds the capacity: collapse at the jump, 7.3 %, closer to the test’s 7.8 %.
span_reference
span increase over the intrados span L = (2R − t) sin α, as in ArchSpread

Modern note

Read in the open-access thesis, chapter 5 and the listing of ArchSpread (Appendix A.1), whose algorithm the engine follows and which it can replay literally: Coulomb’s hinge, rigid-body kinematics of the crown segment, the hinge moving one voussoir when the locus leaves the intrados next to it, the supporting segment taken undeformed after a move, collapse when the locus reaches the extrados below the hinge. The listing counts one voussoir twice after each hinge move; the engine corrects that by default, which explains the one published prediction it did not reproduce (Arch2). By default the hinge may move as far as the locus asks within a step (the converged mechanism); ArchSpread moves it at most one voussoir per step, so with 0.1° voussoirs its hinge lags near collapse and it reports slightly larger spreads (8.2 % instead of 7.8 % for the 120° arch). The journal paper of 2006 was not read.

Sources

Where implemented

@ungewitter/statics#spreadingArch, SpreadTracker, allowableSpreading

Tools in this area

Terms