Microcomputer implementation of simulation benchmarks in heat-transfer and dynamic process analysis

Abstract

In order to foster a better understanding of realistic microcom puter capability, a diverse group of mathematical models, writ ten in FORTRAN, are considered from the standpoint of run ning efficiency. Benchmark results are reported for a variety of computer systems including a mainframe, two minicomputers, and two microcomputers. A description of the simulations, which cover a wide range of complexity, is included. One of the models is a dynamic, two- component material balance of a paper machine. Eleven variable level chests (tanks), a number of intricate controls, and about one hundred stream equations make this model comprehen sive enough to be generally associated with a mainframe, or with very powerful minicomputer systems. Another dynamic simula tion emulates the thermal and stress behavior of flat, injection- molded parts during cooling. It is representative of models based on the one-dimensional, transient heat-conduction equation. Also considered is a steady-state, finite-difference model for prediction of temperature profiles in shell and tube heat ex changers. This simulation employs an efficient, sparse-matrix technique similar to equation solution methods found in finite- element software. Other less comprehensive models, often used as subroutines in process analysis and control, are examined. These include a pump curve versus system resistance search, and a control valve position versus flow computation. Both of these subroutines use iterative solution techniques. The advantages and disadvantages of using the microcomputer are discussed. A range of bounds is established on the feasibil ity of implementing and using a model on a microcomputer in terms of real simulations and run times. The results of this work are very encouraging and indicate a definite place for the microcomputer in simulation work.