Engineering topics
Publications and source records attributed to Salama, M..
This paper presents and examines direct solution algorithms for the linear simultaneous equations that arise when finite element models represent an engineering system. It identifies the mathematical processing of four solution methods and assesses their data processing implications using concurrent processing.
Computer architectures designed for concurrent processing are characterized by the number of processing elements, ensemble speed, random access memory, input/output routes, and modes of operation. The important attributes of processing tasks are then identified, and some processing stratagems are examined. It is shown that the greater the complexity of a given task, the wider the range of possible stratagems which can accomplish the task. For relatively simple tasks, the optimum stratagem can be found by analytical reasoning. For more complex tasks, however, optimum scheduling techniques may have to be employed for the assignment of segments of the task to the available processing elements.
The present study has the objective to investigate some iterative parallel-processor linear equation solving algorithms with respect to efficiency for analyses of typical linear engineering systems. Attention is given to a set of n linear equations, Ku = p, where K = an n x n positive definite, sparsely populated, symmetric matrix, u = an n x 1 vector of unknown responses, and p = an n x 1 vector of prescribed constants. This study is concerned with a hybrid method in which iteration is used to solve the problem, while a direct method is used on the local processor level. Variations in the efficiency of parallel algorithms are explored. Measures of the efficiency are based on computer experiments regarding the algorithms. For all the algorithms, the wall clock time is found to decrease as the number of processors increases.
A mathematical basis is given for comparing the relative merits of various techniques used to reduce the order of large linear and nonlinear dynamics problems during their numerical integration. In such techniques as Guyan-Irons, path derivatives, selected eigenvectors, Ritz vectors, etc., the nth order initial value problem of /y(dot) = f(y) for t greater than 0, y(0) given/ is typically reduced to the mth order (m is much less than n) problem of /z(dot) = g(z) for t greater than 0, z(0) given/ by the transformation y = Pz where P changes from technique to technique. This paper gives an explicit approximate expression for the reduction error e-i in terms of P and the Jacobian of f. It is shown that: (a) reduction techniques are more accurate when the time rate of change of the response y is relatively small; (b) the change in response between two successive stations contributes to the errors at future stations after the change in response is transformed by a filtering matrix H, defined in terms of P; (c) the error committed at a station propagates to future stations by a mixing and scaling matrix G, defined in terms of P, Jacobian and of f, and time increment h. The paper discusses the conditions under which the reduction errors may be minimized and gives guidelines for selecting the reduction basis vector, i.e., the columns of P.
Previously cited in issue 12, p. 1708, Accession no. A83-29871
A computer code, SCOPE, has been developed for predicting the computer resources required for a given analysis code, computer hardware, and structural problem. The cost of running the code is a small fraction (about 3 percent) of the cost of performing the actual analysis. However, its accuracy in predicting the CPU and I/O resources depends intrinsically on the accuracy of calibration data that must be developed once for the computer hardware and the finite element analysis code of interest. Testing of the SCOPE code on the AMDAHL 470 V/8 computer and the ELAS finite element analysis program indicated small I/O errors (3.2 percent), larger CPU errors (17.8 percent), and negligible total errors (1.5 percent).
The results of numerical experiments designed to illustrate how the minimum weight design, accuracy, and cost can be influenced by: (1) refinement of the finite element analysis model and associated load path problems, and (2) refinement of the design variable linking model are examined. The numerical experiments range from simple structures where the modelling decisions are relatively obvious and less costly to the more complex structures where such decisions are less obvious and more costly. All numerical experiments used employ the dual formulation in ACCESS-3 computer program. Guidelines are suggested for creating analysis and design models that predict a minimum weight structure with greater accuracy and less cost. These guidelines can be useful in an interactive optimization environment and in the design of heuristic rules for the development of knowledge-based expert optimization systems.
This paper describes a parallel processing computer consisting of a 16-bit microcomputer as a master processor which controls and coordinates the activities of 8086/8087 VLSI chip set slave processors working in parallel. The hardware is inexpensive and can be flexibly configured and programmed to perform various functions. This makes it a useful research tool for the development of, and experimentation with parallel mathematical algorithms. Application of the hardware to computational tasks involved in the finite element analysis method is demonstrated by the generation and assembly of beam finite element stiffness matrices. A number of possible schemes for the implementation of N-elements on N- or n-processors (N is greater than n) are described, and the speedup factors of their time consumption are determined as a function of the number of available parallel processors.
The paper examines several parallel processing solution algorithms for finite element equations arising in linear equilibrium problems. Two basic groups of algorithms, direct and iterative, are investigated with respect to a number of parallel computer architectures and associated selection criteria. The direct algorithms include: LR-Gauss, Crout, Cholesky, Cyclic Reduction and WZ-factorization. The iterative methods examined are: Accelerated Gauss-Seidel, Surrogate Stiffness, Jacobi, Series Expansion, and Energy Monte Carlo. For real-time applications, where the object is to minimize the execution time, Cyclic Reduction appears to be best suited. This assumes a computer with an unlimited number of parallel processors. However, for computers with a limited number of parallel processors that must be used efficiently, both Gauss factorization and Jacobi-like iterative methods rank favorably.
The nonlinear strain-displacement and velocity-displacement relations of spinning imperfect rotational paraboloidal thin shell antennas are derived for nonaxisymmetrical deformations. Using these relations with the admissible trial functions in the principle functional of dynamics, the nonlinear equations of stress inducing motion are expressed in the form of a set of quasi-linear ordinary differential equations of the undetermined functions by means of the Rayleigh-Ritz procedure. These equations include all nonlinear terms up to and including the third degree. Explicit expressions are given for the coefficient matrices appearing in these equations. Both translational and rotational off-sets of the axis of revolution (and also the apex point of the paraboloid) with respect to the spin axis are considered. Although the material of the antenna is assumed linearly elastic, it can be anisotropic.
It is noted that the generalized modal shock spectra method had limitations on the degree of structural static determinacy of the spacecraft-to-launch-vehicle interface. The 'interface modes' is used here to remove these practical limitations. The governing differential equations are derived and then shown to be valid either for integrating the modal models of two or more substructures that have been previously obtained separately or for removing a substructure from a previously available system modal model. Emphasis is placed on the integration problems in the context of the generalized modal shock spectra approach. A minimum of information the modal response of a previously obtained transient analysis of the unloaded launch vehicle is exploited in order to define an idealized modal impulse function and derive approximate explicit expressions for an estimate of the bounds of the spacecraft response. In addition, a numerical model is solved in order to compare the present results with the transient solution.
A finite element analysis emulator providing a basis for efficiently establishing an optimum computer implementation strategy when many calculations are involved is described. The SCOPE emulator determines computer resources required as a function of the structural model, structural load-deflection equation characteristics, the storage allocation plan, and computer hardware capabilities. Thereby, it provides data for trading analysis implementation options to arrive at a best strategy. The models contained in SCOPE lead to micro-operation computer counts of each finite element operation as well as overall computer resource cost estimates. Application of SCOPE to the Memphis-Arkansas bridge analysis provides measures of the accuracy of resource assessments. Data indicate that predictions are within 17.3 percent for calculation times and within 3.2 percent for peripheral storage resources for the ELAS code.
The second order nonlinear equations of motion are formulated for spinning line elements having little or no intrinsic structural stiffness. The derivation is based on the extended Hamilton's principle and includes the effect of initial geometric imperfections (axial, curvature, and twist) on the line element dynamics. For comparison with previous work, the nonlinear equations are reduced to a linearized form frequently found in the literature. The comparison revealed several new spin-stiffening terms that have not been previously identified and/or retained. They combine geometric imperfections, rotary inertia, Coriolis, and gyroscopic terms.
Numerical solution of nonlinear equilibrium problems of structures by means of Newton-Raphson type iterations is reviewed. Each step of the iteration is shown to correspond to the solution of a linear problem, therefore the feasibility of the finite element method for nonlinear analysis is established. Organization and flow of data for various types of digital computers, such as single-processor/single-level memory, single-processor/two-level-memory, vector-processor/two-level-memory, and parallel-processors, with and without sub-structuring (i.e. partitioning) are given. The effect of the relative costs of computation, memory and data transfer on substructuring is shown. The idea of assigning comparable size substructures to parallel processors is exploited. Under Cholesky type factorization schemes, the efficiency of parallel processing is shown to decrease due to the occasional shared data, just as that due to the shared facilities.
In this paper, developments in digital computer technology since the early Fifties are reviewed briefly, and the parallelism which exists between these developments and developments in analysis and design procedures of structural engineering is identified. The recent trends in digital computer technology are examined in order to establish the fact that distributed processing is now an accepted philosophy for further developments. The impact of this on the analysis and design practices of structural engineering is assessed by first examining these practices from a data processing standpoint to identify the key operations and data bases, and then fitting them to the characteristics of distributed processing. The merits and drawbacks of the present philosophy in educating structural engineers are discussed and projections are made for the industry-academia relations in the distributed processing environment of structural analysis and design. An ongoing experiment of distributed computing in a university environment is described.
General nonlinear discretized governing equations of motion of spinning elastic solids and structures are adjusted for the case of a spinning cantilever with initial geometric imperfections. Consideration is given to second degree nonlinearities in the strain-displacement and velocity-displacement relationships. Parameters of the discretization are developed to include the type and number of the coordinate functions used in the admissible trial solution in order to unify the discretization approaches associated with stationarity principles. The coordinate functions comprise both sets of continuous and piecewise continuous functions employed in the Rayleigh-Ritz and the finite element methods, respectively. Coefficient matrices are provided which contain the energy density expressions and which are adaptable to computer programming.
The structural-synthesis computer code ACCESS-3, originally limited to the optimization of relatively simple problems involving truss, membrane, and shear panel elements, has been restructured to make it applicable to structures with a larger number of degrees-of-freedom, design variables, and constraints. The modifications made allow the use of six degrees-of-freedom per node and permit the out-of-core solution of large matrix equations encountered at the analysis stage. Also, finite elements which permit pure bending as well as combined bending and membrane (axial) behavior have been added to satisfy the need for a more comprehensive set of elements. Like its predecessor, the new version of the program combines the approximation concepts with primal or dual formulations to provide an efficient optimization tool.
The general nonlinear discretized equations of motion of spinning elastic solids and structures are derived as a set of nonlinear ordinary differential equations for the case when the strain-displacement and velocity-displacement relations are nonlinear up to the second order. It is shown that the cost of generation of such equations is proportional to the fourth power of the number of degrees of freedom. A computer program is written to automatically generate the equations for the case of spinning cantilevers with initial imperfections. The types and the number of the coordinate functions used in the trial solution are parameters of the program.