RuleModern statics
The arch on leaning buttresses
modern.kinematics.arch-on-leaning-buttresses
Stated in a medieval text.
Statement
As its buttresses lean, an arch is pulled wider: its thrust rises while the buttresses’ capacity falls. One of two things ends it. If the buttresses are strong, the arch reaches the greatest spreading it can follow and collapses by its own mechanism, the buttresses intact (strong-buttress failure); this is the usual case for thin arches. If the arch is thick, its thrust can overtake the falling capacity first: the buttress gives way, rotates further, and the arch falls (weak-buttress failure). Either way the buttress usually remains standing, relieved of the thrust. Tall, slender buttresses lift the springing, so a small lean widens the span more and their capacity falls faster: such arches can fall at a lean under 1°.
Formula
Parameters
- validation_goa.arch_R_m
- 5
- validation_goa.t_over_R
- 0.1
- validation_goa.half_angle_deg
- 60
- validation_goa.leaning
- south wall only
- validation_goa.thrust_at_0_4_deg_kN
- 41
- validation_goa.collapse_lean_deg
- slightly more than 2.0
- validation_goa.mode
- weak-buttress
- validation_goa.thrust_at_collapse_kN
- 52
- validation_goa.crown_drop_m
- 0.4
- validation_goa.thin_arch.t_over_R
- 0.05
- validation_goa.thin_arch.collapse_lean_deg
- 1.1
- validation_goa.thin_arch.mode
- strong-buttress
- validation_goa.source_page
- 131–134
- engine_goa
- 2.01°, weak-buttress, 52.3 kN, crown down 0.41 m, 41.5 kN at 0.4°; t/R = 0.05: 1.09°, strong-buttress
- case_a_fig_7_3
- collapse leans 0.8°, 2.1°, 3.4°, 4.75° for t/R 0.05–0.20 (read off the figure); engine 0.76°, 2.06°, 3.36°, 4.71°
- threshold_fig_7_5
- Reproduced by ArchLean’s own stepping (archOnLeaningButtresses with listing: true): the first weak-buttress failure at t/R between 0.155 and 0.1575 for Case A (printed 0.156) and between 0.103 and 0.1045 for Case B (printed 0.104). Two features of the listing decide it (pp. 166–169): in each step the buttress is checked before the arch, and right after a hinge move the thrust is computed with the crown segment’s centroid unrotated (XM1 = X_O + C1 sin(p/2), where the main loop has sin(p/2 − φ)), which understates it, so the arch survives the jump and in the next step the overshooting thrust is charged to the buttress. With the centroid rotated the listing’s thresholds are 0.171 and 0.107.
- threshold_engine
- The engine follows the converged mechanism and, where a hinge jump overshoots both limits at once, lets the weaker part give way: Case B changes mode between 0.1045 and 0.106, Case A between 0.165 and 0.1675. Above that the arch’s own collapse and the buttress’s giving way fall within one hinge step of each other, and with 1° voussoirs the mode alternates with t/R (for Case A strong again at 0.1975–0.2, 0.21–0.215, …) while the lean at collapse runs on smoothly (about 0.068° per 0.0025 of t/R). The threshold thickness is therefore ill-conditioned; the collapse lean is not. Until wave 6 the engine charged every simultaneous overshoot to the arch, which put the thresholds at about 0.117 (B) and above 0.2 (A).
Modern note
Read in the thesis, §7.3 and §8.2, with the listing of ArchLean (Appendix A.3), which the engine can also replay statement by statement: the spreading arch of chapter 5 on the cracked rectangular buttress of chapters 3–4, the buttress capacity reduced by eq. 4.9 as it leans, both buttresses leaning (or, at Goa, only one: the span then grows by h·φ instead of 2h·φ).
Sources
Ochsendorf 2002, p. 112–121, 129–134, 164–170, Fig. 7.2–7.6, 8.1–8.3, App. A.3
Where implemented
@ungewitter/statics#archOnLeaningButtresses