In twiss for the SBEND I found around line 3919 the following equation in the radiation (hx2 + hy2) * (one + h*x)
The factor (one + hx) = h(rho+x) is scaling the arc path-length in the Hamiltonian (for a constant angle theta) so when applied to curvature h = dtheta / ds = 1 / rho, I think that it should divide it (not multiply it).
We can also look at the variation of the above quantity vs x, i.e. increasing x means increasing the curvature, but it should decrease because the path-length increases.
In twiss for the SBEND I found around line 3919 the following equation in the radiation (hx2 + hy2) * (one + h*x)
The factor (one + hx) = h(rho+x) is scaling the arc path-length in the Hamiltonian (for a constant angle theta) so when applied to curvature h = dtheta / ds = 1 / rho, I think that it should divide it (not multiply it).
We can also look at the variation of the above quantity vs x, i.e. increasing x means increasing the curvature, but it should decrease because the path-length increases.