Exact controllability of a thermoelastic system with control in the thermal component only (Q5933778)

The problems of controllability, stability, observability of controlled thermoelastic systems have been subject to much work, both by engineers and mathematicians. In particular controls applied (realistically) to the boundary have been of much interest. This paper is a continuation of the works of the author in cooperation with Irena Lasiecka, and those of Lasiecka and Triggiani. If we equate most of the physical coefficients to one, modelling the free vibration of a thin plate coupled with heat transfer and using the Kirchhoff plate model and Fourier's law of heat transfer, we may derive the author's system: \[ \omega_{tt}- \gamma\Delta\omega_{tt}+ \Delta^2\omega+ \alpha\Delta\theta= 0,\tag{1a} \] \[ \theta_t- \Delta\theta+ \sigma\theta- \alpha\Delta\widetilde\omega_t= 0.\tag{1b} \] If a control consists of a thermal equation damping only, the zero on the right-hand side of (1b) is replaced by a control \(u\in \{L^2(0, T)\times H^{-1}(\Omega)\}\). Here \(\omega\) is the transverse displacement of the plate, \(\theta\) is the temperature. The initial temperature and displacement as well as the initial velocity distribution are known. \(\alpha\) is the coupling parameter. The parameter \(\gamma\) is crucial because it determines the nature of the differential system. If \(\gamma> 0\), then the system is hyperbolic-like, while if \(\gamma= 0\) then the generating semigroup is analytic, thus producing parabolic type of dynamics. In the hyperbolic case it has been shown in past papers of Lasiecka and the author that it is possible to obtain exact controllability on displacement, and partial controllability on temperature, without assuming that \(\alpha\) is small. Some results were obtained recently by uncoupling the displacement from temperature, and then regarding the present system as a perturbation of the ``pure'' Kirchhoff model. The author does better by proving the following theorem. For all \(\gamma>0\), for any \(T>0\), and any data in the appropriate space \(\{L^2(\Omega)\times H^1_{0,\gamma}(\Omega)\times H^2_0(\Omega)\}\), one can find an exact control. He comments on the full consequences of this result. For example, the displacement can be exactly controlled by placing a control consisting of a heat equation input over an arbitrarily short time interval. This seems to be too good to be true. In a way it is. While the author's result is important in mathematical studies of equations of the type (1a), (1b), and of similar systems, it has to be viewed with suspicion by an engineer, trying to control a vibration say of a circular plate with 100:1 ratio of the radius to the (constant) thickness by heat flux control placed in the heat transfer equation. Propagation of heat transfer, modelled in practical cases by Monte Carlo (random walk) techniques is of finite speed, increasing with the square root of the local averages of absolute temperature. It appears after a rough estimate that to affect a rapid reduction of vibration by changes in local temperatures one shall need temperatures too close to absolute zero, or too close to the temperature of the sun, to be generated. Never mind the fact that Kirchhoff's model works in a very limited temperature range for elastic materials, such as medium carbon steel, even if it contains some exotic additives. The reviewer voices a concern on the applicability of many such results. Kirchhoff, Timoshenko, Vlasov's models, try to describe real plates, shells, etc\dots The Mindlin or Kirchhoff models of plates are studied in such detail because they describe closely behavior of plates in the elastic range of the material. Also there is a great variety of fourth-order partial differential equations which we could, but do not bother to study, because there is no demand from other scientist or engineers to do so. And, by the way, the name of that great scientist is Gustav Kirchhoff, not Kirchoff.