applications

Potential for micros in process by PIERRE TRAN

lthough microcomputers have

A found widespread acceptance and use in the business, profes-

sional , educational and hobbyist sectors, their penetration of the

engineering professions has been much slower. In certain engineering disciplines, relatively complex pro-

grams are run on computers, but the process engineer has not, up till now, fully exploited the practical possibili- ties of the micro in everyday use.

The process, or more specifically, the chemical engineer is a key pro- fessional in a range of industrial sectors which include oil, chemicals,

fibres, pharmaceuticals, steel and food processing, all of which draw heavily on computing products and

power. Such is the complexity of the design, engineering, operation and

support functions which make up the responsibilities of the chemical

Abstract: Microcomputers offer considerable cost savings in process control, particularly for local data entry and validation. Among the applications which can be performed on a microcomputer include chemical process simulation utilizing historic data, as well as modelling unit operations, equipment design and rating and process optimization and integration. The character of chemical engineering projects are discussed to illustrate how they are suited to limited memo y microcomputers rather than large mainframes.

Keywords: data processing, microcomputers, process control.

Pierre Tran was consultant at the Economist Intelligence Unit and ia, now a technical journalist.

control

engineer, that it is inevitable that he/she will account for a large portion

of the corporate computing bill. It is for this reason that the potential of microcomputers should be examined

more closely. Microcomputers com- bine low cost, reasonable power, ease of use and relative portability.

Cost advantages of using micros for process control

The cost advantages of introducing an inhouse micro to run process control

programs compared to time-sharing on a mainframe, should be recog-

nized. In one company studied as part of a review conducted by the Eco- nomist Intelligence Unit’, nearly $60000 was spent over a period of

months on agency computing. In- house developed reservoir and linear

programming models were used. It was found that data storage and connect time accounted for over 20% of costs, and computer resource units for most of the rest. Even if all the

calculations and processing were still done on the remote computer, the savings made by using a micro for local data entry and validation were

shown to cost justify its $5 000 price in approximately eight months. This cost would also include a high-speed

matrix printer, enabling the system to function as a discrete intelligent terminal. A considerable amount of local processing and data storage would also be possible. A relatively simple 8 bit machine was used, there- fore, it would be likely that greater savings might be achieved by using more powerful and versatile 16 and 32 bit micros.

The type of unit operations which are common to a variety of process

plants and which can be run on a microcomputer include:

separation operations such as dis-

tillation, flash separation, crystalli- zation, evaporation, drying, ab- sorption and liltration;

physical transport or mass transfer operations such as the movement of liquids and slurries through pipelines by pumping, the move-

ment of gase:s by compression or the movement of solids by con- veyors; heat transfer operations including

heat exchange between process streams, cooling of process streams by air, water, or some other

medium, heating of process

streams in fired heaters.that typic-

ally burn low-value gas, oil or coal to generate the heat required;

The technical, economic and manage- ment activities which confront the chemical engineer are all extensively

supported by computer-based tools. The following describes three of the main types of programs used in these activities.

Steady state process simulation

This is commonly used by oil and chemical companies to generate heat and material balances across process

plant systems by performing a simu- lation of the plant under steady state operating conditions. Such simulators offer the capability of modelling virtu- ally every conceivable continuous pro- cess designed in the process indus- tries.

~0125 no 10 december 1983 0011-684Xi831100035-02$03.00 @) 1983 Butterworth & Co (Publishers) Ltd 35

Chemical process simulation

This type of application is mainly

used by operating companies, con-

tractors, and process licensers to pre- dict yields, qualities and other para-

meters, such as reactor run lengths, for several major refining and petro- chemical processes. The programs are used both at the design stage and to

monitor the operation of installed plants. They are most widely used in catalytic reforming, and catalytic,

hydro or steam cracking. These pro- grams differ from those described in

the previous section in that they not

only model unit operations but are more likely to work from historic data from past operations, test runs or

equations which simulate the reaction kinetics of a process.

Process control and online computing

This is used to log data from plant instrumentation, generate control sig- nals to correct plant conditions or

hard copy analysis for management.

Other programs which could be run on a microcomputer are:

dynamic simulation equipment design and rating process optimization and integra-

tion energy management engineering drawing capital and operating cost estima-

tion operations and investment plan-

ning and scheduling

project engineering and manage- ment.

Characteristics of chemical engineering programs

The main types of programs used by the chemical engineer cover a wide range and their computing character- istics vary accordingly. Many of the technical programs used are compara- tively straightforward routines involv- ing single pass calculation of a series of formulae or equations. Their data requirements are often restricted to a

limited set of numbers input by the

user, such as stream starting tempera- tures and pressures at the surface

area, and heat transfer coefficients of a

heat exchanger. This class of program includes most of the simpler unit operation calculations and chemical

process simulators, heat exchanger and fired heater rating and design programs, single line pressure drop calculations and energy management

tools. The comprehensive simulation

packages and the linear program opti-

mizers make sizeable demands on CPU power. A successful simulator

needs to have a comprehensive data- base, with a fast and accurate calcu-

lating capability and access, typically 2 to 3 megabytes of memory.

Simulations that involve multiple

processes, distillation columns and recycle streams will become heavily interative, repeating calculations of individual unit operations many times

to arrive at a converged solution. The CPU time required for running an

extensive simulation on a large main- frame may vary from 2 to 30 minutes.

Although most leading commercial LP systems can advantageously use large amounts of main memory, they will function, albeit more slowly, if allocated only around 100 kbyte.

These optimizers are built up from a number of complex mathematical al- gorithms, may make heavy use of work file space during execution and

will consequently impose a substantial I/O burden on the system.

Simple LP systems can be written directly using published equations and these will solve problems of approximately 100 rows. They oper- ate using an array that must be stored in main memory, and since every coefficient (including zeros) in the matrix is presented as an array in element, the maximum problem size that can be accommodated is limited by the available main memory or the amount which can be addressed by the virtual memory system. To accommodate a typical medium-sized

model, such as a 300 row refinery matrix, the computer will need to

access at least 128kbyte of main or

virtual memory. To summarize, the major charac-

teristic of chemical engineering pro-

grams, is that the majority of memory and CPU requirements can be ade- quately handled on low to medium

power micros and with limited amounts of memory. It is the large simulators and LP systems which need main memory and the pro-

cessing speed of mainframes. Existing chemical engineering computer tools

have been highly developed, but

their user features and computing characteristics reflect early software and hardware, and are, therefore, rooted in past approaches to problem-

solving. Tomorrow’s computer-based tools are likely to be different, given the availability of micro technology.

The main hardware requirements

which will affect the long-term poten- tial of microcomputers in running applications requiring high perform-

ance are:

processors which can handle high data rates in conjunction with high computational speeds

a system architecture that supports multiple processors a bus structure with wide band-

width intelligent, buffered peripheral

controllers a large amount of main memory

The current generation of micros already provides a useful tool for

process control engineers but it is the micros of the mid to late 198Os, based on 32 bit architecture, that will rival the tasks performed by today’s mainframes.

Reference

1 Economist Intelligence Unit, Micro- computers - their technical and market potential in the process industries, EIU, London (1983). 0

Economist Intelligence Unir Ltd, Spencer House, 27 St. James’s Place, London SWlA INT, UK. Tel: 01493 6711.

36 data processing