A stay cable is tensioned by the bridge weight; engineers estimate that tension from the cable's measured vibration frequency. The pluck excites the first in-plane modes of a finite-difference cable (48 intervals, leapfrog time step) that includes bending stiffness and the elastic extra tension caused by sag. A virtual accelerometer at mid-span counts zero crossings to get f1, and three ways of turning f1 into tension are compared with the true value you set.
m y_tt = T y_xx - EI y_xxxx + h (w/T) , h = -(EA w / (T L)) * integral(y dx) (w = m g cos(theta))
taut string: f1 = (1/2L) sqrt(T/m) => T = 4 m L^2 f1^2
beam-string: f_n = (n/2L) sqrt(T/m) sqrt(1 + n^2 pi^2 EI / (T L^2))
Irvine sag: tan(x) = x - 4 x^3 / lambda^2 , x = omega L sqrt(m/T)/2 , lambda^2 = (w L / T)^2 EA / T
- What to try - "Short (bending)": the EI term raises f1 and the naive formula overestimates T. "Long (sag)": lambda^2 grows and the first symmetric mode is stiffened by the cable's extra tension, so the naive estimate is far too high. Raise EI or EA, change inclination (smaller cos theta means less sag) or lower T and watch the errors.
- Checks - FD against the beam-string closed form, FD against the Irvine transcendental root, the sag-free antisymmetric mode 2, second-order grid convergence, undamped energy and the tension identification error; all run live on your parameters.
- Same model in the 2D twin - one physics module; the 3D page and the elevation view only draw it. Cable displacement is drawn magnified (display control).
- Honest simplifications - hinged ends (real stays have fixed sockets, a guide pipe and often a damper, which shift the effective length); linear small-amplitude in-plane vibration; parabolic static profile; chord length instead of the curved length in lambda^2; sag and bending combined by adding the two frequency effects (an approximation, good to a few hundredths of a percent for the presets). Teaching model, not a cable-monitoring tool.