Numerical approximation of the dynamic Koiter's model for the hyperbolic parabolic shell (Q2301289)

The aim of the study remains a bit unclear to the reviewer. The analytical background of the FEM style of smooth shell analysis, with the negative Gaussian curvature ($K<0$), has been done during the 60s--80s of the past century in a very advanced form (static/dynamic, linear/nonlinear, etc.). Rotary hyperboloids (cooling towers), hyperbolic-paraboloids (roofs), various surfaces of straight lines, etc. Tensor style of analysis has been predominantly involved in order to avoid unpleasant information gaps in curvilinear space. The differential geometry of thin-walled shells introduced in Section 2 is a standard tool, which is presented in many monographs and papers since the 1950s. The linear approximation with respect to middle surface introduced in the beginning of Section 3 is an approach widely used everywhere, when investigating static/dynamic problems of general thin-walled shells. Moreover, linear approaches have been intensively studied in the 1960s and 1970s using various polynomial expansions in $\xi$ variable (e.g., Reissner, Bloch, and others ) to admit higher thickness of the shell. Another approach appeared using an asymptotic expansion (in invariant formulation) with respect to normal $a_3(y)$, e.g., Kilchevskij, Goldenweiser, etc., see monographs and papers from the period 1965--1990). The quadratic functional (3.1) in many variants has been discussed many time for isotropic as well as anisotropic materials and more complicated formulations (nonlinear approaches geometric/physical are not commented in this review). Time discretization following the Newmark scheme, as presented in the paper, Section 4.2, belongs to very standard tools including a number of modifications intended for special purposes. It seems that the Newmark scheme is satisfactory, provided the surface is smooth enough (more sophisticated procedures, like gear, etc., can be avoided). Nevertheless, the dynamic character of the system is not analyzed carefully. The excitation function considered in the paper is very simple in time and cannot lead to dynamic effects, which can bring the stability loss (structural/or numerical) or lead to other dangerous phenomena. Nothing about eigenvalues and eigenvectors is reported, where the stability is usually fragile. How about the frequency upper limit with respect to the size of the largest element, etc.? Convergence of the numerical process and solution unicity in the domain of discretized continuum as characterized by Theorems 4.1 and 4.2 have been discussed many times in much more general assignment. On the other hand, it is true that existing books and articles are very schematic and difficult to study for a reader who is not proficient in tensor algebra and analysis. From this point of view, the article is very useful not only in the field of shell theory but all the tasks of field theory in curvilinear and non-orthogonal coordinates. A detailed example (Valencia aquarium) of the analysis gives clear answers to many questions that the reader may encounter in the interpretation of the interplay of covariant and contravariant frames together with physical coordinates. Also numerical analysis in a very concrete case is a useful demonstration of practical applicability. Moreover, the results published in the paper can be easily verified using commercial packages (ANSYS, ABAQUS, etc.).