| dc.description.abstract | Earthquake safety engineering of nuclear power plant structures poses several challenges to the analyst and designer. These problems are characterized by highly transient and dynamic nature of earthquake induced excitations, random nature of details of support motions (in terms of duration, frequency content, amplitude modulation, multiple components, and spatial variability), nonlinear nature of structural behavior, geometrical complexity of the primary and a large number of secondary systems (such as, for example, piping, rotors, and machine panels), soil-structure interactions, demands on high level of safety expected of these structures, and general paucity of recorded data on strong ground motions appropriate for the given site. Probabilistic methods offer the most rational framework to base design decisions for this class of problems. The work reported in the present thesis belongs to this broad area of research. We focus attention on studying two classes of nuclear power plant components, namely, a pipework in the heat exchanger segment, and, control and safety rod drive mechanism (CSRDM) and investigate their performance by taking into account complicating features such as differential seismic support motions across multiple supports, nonlinearities at support locations, random nature of dynamic loads and uncertainties in system parameters. Response measures include peak responses, reliability against specified performance criterion, measures of uncertainties in response variables of interest.
Chapter-1 provides the functional details of nuclear power plant structures that includes reactor assembly and heat transport system assembly, CSRDM, heat transfer piping networks, and nonlinear supporting devices (such as rod, spring, guide supports, limiters, and snubbers). The discussion brings out the structural mechanics issues that need attention while analyzing seismic response of some of these components. Chapter-2 provides a brief review of literature covering the following topics: Monte Carlo simulation based methods for static and dynamic reliability analysis problems, digital simulation of random variables and processes, treatment of non-Gaussianity in simulations, strategies for variance reduction, models for uncertainty in response using limited samples, data based extreme value analysis, studies on multi-supported piping networks under differential seismic inputs and seismic performance of CRDM structures. The study identifies specific issues related to numerical simulation of nonlinear dynamic response of multisupported pipeworks to differential seismic inputs, uncertainty propagation and reliability modeling in seismic response of pipeworks and CSRDM using Monte Carlo simulations with variance reduction, data based extreme value analysis, and uncertainty propagation using limited samples as topics requiring further research.
The problems of numerical simulation of nonlinear multisupported piping systems subjected to differential seismic support motions and drop time characterization of CSRDM structure during a seismic event are considered in Chapter-3. It is noted that commercially available professional finite element analysis (FEA) softwares do not offer a direct means to tackle this class of problems. On the other hand, FEA packages are best suited to produce acceptable FE models which take into account the geometrical complexities of the structures. Thus, the reasonable way to move forward would be to develop external interfaces that take advantage of FE modeling capabilities of professional packages and at the same time enable treatment of complexities associated with differential support motions, nonlinearities and axial rigid motions of subsystems as in CSRDM. The work reported in Chapter-3 describes the efforts expended in achieving this objective. Here the given built-up structure is divided in to a set of linear substructures each of which are modeled using FE analysis procedures. The proposed scheme allows for these FE models to reside in professional FE analysis codes. An iterative time domain scheme for modeling the interaction forces between these substructures is discussed. The set of governing equations of motion are developed in terms of normal modes of substructures in their uncoupled states. A suite of benchmark problems are first employed to validate the procedure developed. Subsequently, the earthquake induced dynamic response of CSRDM structure and the pipeline running between IHX and secondary sodium pump in a typical fast breeder reactor is simulated.
The algorithm for simulation of dynamic response of CSRDM and multi-supported pipelines under differential support motions developed in Chapter-3 is employed in Chapter-4 to investigate the questions concerning influence of uncertainties in specifying the loads and the system parameters on the system response. Specifically, the study focuses on quantifying uncertainty in system response characteristics based on limited number of Monte Carlo simulations of the response. For this purpose we draw upon an earlier work by Wilks which specifies the number of samples needed to estimate γ th percentile point of a random variable with β level of confidence. We explore in this
Chapter, the application of this idea in the analysis of nonlinear, randomly parametered, dynamical systems under stochastic excitations.
In Chapter-5 we turn our attention to the modeling of aseismic reliability of the nonlinear pipework under differential support motions and the CSRDM structure. The performance functions considered for the piping structure are in terms of highest displacements and stresses over a specified time durations while for CSRDM, the performance function is in terms of scram time being less than a specified time duration. We tackle the first problem by using theory of data based extreme value analysis while the second problem is addressed using an adaptive importance sampling strategy. The contributions here pertain to the exploration of data based extreme values analysis as applied to an industrial scale structure and improvisation of algorithmic modifications in the development of adaptive importance sampling density functions. This improvisation consists of selection of sampling points as a judicious mix of points from both safe and unsafe regions. This is shown to reduce the strong correlations that otherwise would be present if samples are taken only from the unsafe region. These studies demonstrate how Monte Carlo simulations with limited samples can be utilized to draw useful conclusions on structural reliability.
Chapter-6 summarizes the main contributions made in the thesis and makes a few suggestions for further research.
There are five annexures in the thesis. Annexure-1 contains listing of Matlab m-files used for solving illustrative problems in Chapter-2. The details of FE modeling of multisupported system under differential support motions and the details of substructuring scheme used in modeling of such systems with local nonlinearities are provide in Annexure-2. The details of material and geometry of CSRDM structure are provided in Annexure-3. Annexure-4 summarizes the main details of hypothesis tests used in data based extreme value analysis. The algorithms used for converting response spectra into compatible power spectral density functions are described in Annexure-5. | en_US |
