Advances in Engineering Software (1978)最新文献

KBase is a customizable development tool for building knowledge-based systems that are compatible with DBase III tables. Currently, most tools for developing knowledge-based systems do not integrate well with databases. DBase III is the most common database system used in the personal computer environment. KBase is built such that all DBase III commands and functions are supported. Furthermore, customizing and extending KBase is as easy as programming in DBase. This flexibility arises because KBase is built using the “C” programming language and Clipper, a DBase compiler. Using Clipper also allows KBase to use many functions in the Clipper library. In this paper, we describe how KBase's design is conceived. At the end, we also give a brief tour of the program with an implemented example to illustrate its application.

A suite of FORTRAN 77 computer programs implementing Bayesian nonlinear regression is described. These programs, developed to cope with complex hydrologic models, are used to infer model parameters, test model structure and make predictions of future response. The programs are fully interactive with features including: interactive parameter optimization using the Gauss-Marquardt algorithm; run-time editing of user options; availability of a general error model; use of prior information; and joint fitting of multiple-response data.

This article presents a non-linear transient thermal analysis finite element program which is developed based on the goal in simulating the casting process with the phenomena of solidification and changes of latent heat.

The continuous casting problem was experimentally investigated in 1953 by Brandt, Bishop and Pellini¹, who used thermocouples in the mold and mold cavity. The experimentally measured isochronal solidification profiles for low carbon steel cast in simple ‘T’ shaped sand molds were matched by calculations from the finite element program, with appropriate physical assumptions. The computations from the heat transfer model agreed with the experimental results with respect to the shape of the isochronal solidification profiles, time of solidification, thermal gradients and temperature profiles.

A simple algorithm is presented for removing the hidden lines from an object which consists of a collection of solid blocks. The elementary blocks are general convex polyhedra with any number of faces. The representation of the object as a whole requires that the lines of separation between contiguous blocks are also removed.

The procedures developed are described in detail and illustrated by flowcharts.

A unified numerical statement, which composes Galerkin finite element, subdomain finite element, finite control-volume, balanced finite difference and other numerical approximations for 2-D and 3-D axisymmetric CϖT/ϖt = Δ · (KϖT) + λT + F, was derived and formulated. The numerical statement yields Galerkin finite element, subdomain finite element, and balanced finite difference approximations by specifying a single constant, η, to 2,3 and ∞ respectively (finite control-volume approximation by lumping capacitance matrix of the subdomain finite element method). Techniques are also presented to illustrate selection of input parameters to UniSub, a Fortran implementation of the unified numerical statement, to generate correct matrix systems for the governing equation and various boundary conditions. Test problems were run to check the validity of the subroutine and to compare accuracy of various numerical schemes.

UniSub is a powerful tool to compare the accuracy of various numerical schemes since it eliminates the uncertainty of effects between codes used for comparison. UniSub can also be employed in a Fortran program to produce different numerical schemes by varying η both spatially and temporarily to achieve optimal accuracy in solving diffusion equation.

The present work describes a method for the determination of the position of a point with respect to a simple-connected domain. It is intended for application in the finite element mesh generation. The proposed method for identification of a point with respect to a domain appears as an extension of Sloan's method, which is an improved version of Nordbeck's and Rydsteadt's one, for identification of a point with respect to a polygon. FORTRAN 77 computer programs are added implementing the proposed method.

It is frequently necessary to manipulate large sparse matrices by means of a computer. In such cases a lot of CPU time and memory space can be saved if only the non-zero elements are stored. This paper surveys seven different “compact” representations of sparse matrices. The selected implementations will be compared with regard to the running time and the storage requirement.

This paper describes the development of a computer software for analog to digital data acquisition. The major system hardwares consist of an IBM/AT compatible together with a Metrabyte Dash 16/F and 8 Exp-16 interface boards. The system is capable of providing 128-channel on-line data acquisition at a maximum sampling rate of 0.2 Hz. The raw data from all channels are simultaneously monitored on a screen in numerical form. Upon completion of data acquisition, the system allows the user to display the raw data stored in a data file graphically for each channel. The system is versatile and economical and is ideally suitable for laboratory work. Program listings are attached as a reference base for further work development. It is suggested to use block channel scan technique to increase the data sampling rate.

This paper presents a Fortran implementation of the Fast Approximate Inverse (FAPIN) algorithm for solving certain large scale sparse linear systems in a triangular domain developed in ³. The algorithm can be used to solve very large linear systems that arise from the finite element solution to several types of engineering and applied science problems. A brief description of the implementation is given and a simple example to illustrate the use of the algorithm is also provided.