Decay rates for a beam with pointwise force and moment feedback. (Q1865840)

The authors discuss the dynamic response and feedback behavior of the vibrating homogeneous Rayleigh model and of the Euler-Bernoulli model of a beam subjected to feedback damping forces and moments, which are represented by pointwise loads (Dirac delta) and pointwise moments (the first spatial derivative of the Dirac delta function). The length of the beam is \(\pi\), and consequently \(0<x < \pi\), \(t> 0\). It was shown back in 1966 by the reviewer and later by others that optimal damping involves application of pointwise control loads and moments. This corresponds with the assumptions of the authors. The Rayleigh model is reduced to the form \[ \partial^2u/\partial t^2-\partial^4 u/\partial x^2\partial t^2+ \partial^4 u/\partial x^4+ (\partial u(\xi, t)/\partial t)\circ \delta_\xi- (\partial^2 u/\partial x\partial t)\circ d\delta_\xi/dx= 0, \] with free support at both ends, and specified initial displacement and velocity. The Euler-Bernoulli beam has a similar form, except that now the density \(\rho\) is multiplying the acceleration \(\partial^2 u/\partial t^2\), and \(\text{EI}(x)\) reduced to a single letter \(p(x)\) is inserted into the fourth-order spatial term, in the usual manner. At \(x= \xi\) the authors assume existence of a hinge, where both the control force and the moment are applied. In effect they created a system of two beams connected by a hinge, where they applied the Dirac delta and its first derivative. The main results are energy decay theorems. For the Rayleigh beam the authors show that the decay is exponential if and only if the ratio \(\xi/\pi\) is rational. This strange result seems to be appearing regularly in the literature. The first time the reviewer encountered it was in the late sixties, reading an MRC report, which was later published. I recall a result which turned up in the literature as far back as 1970s. More recently, a very active researcher in control theory, R. Triggiani, derived in 1993 a similar type of result, stating that there exists a dense set of points in the domain where a wave in more than one dimension can be asymptotically stabilized. Here, the authors state that the existence and uniqueness of solutions implies that the energy of the system controlled at the hinge point \(x=\xi\) is exponentially decreasing with exponential constant \(\omega\) depending only on \(\xi\), if and only if \(\xi/\pi\) is a rational number of the form \(p/q\), where \(p\) and \(q\) are relatively prime, and \(q\) is an odd integer. However, in any case polynomial rate of energy decay is proved. This is an interesting theoretical result for the type of equation studied here, but it really has no clear practical applications. For this result to be duplicated in experiments (how?) one cannot start with assumptions leading to the Rayleigh or Euler-Bernoulli, or Timoshenko models, or any other equation that is supposedly modelling real-life phenomena. The word ``beam'' should be deleted in the title, and elsewhere in the paper. One could argue that point force and point moment are also abstractions, impossible to realize. But there is a fundamental difference. The engineers or physicists feel comfortable with these abstractions. It is easy to approximate a point load by concentrating force in a vary small region. A bullet fired from a gun can be considered for practical reasons to be a point mass, when its trajectory is computed. But there is no way to approximate a rational number creating a region supposedly devoid of irrationals, or approximate a rational value of any physical constant. Here we must not tell engineers working in control theory that we have results that they can use.