Citation
Chemical engineering education

Material Information

Title:
Chemical engineering education
Alternate Title:
CEE
Abbreviated Title:
Chem. eng. educ.
Creator:
American Society for Engineering Education -- Chemical Engineering Division
Publisher:
Chemical Engineering Division, American Society for Engineering Education
Creation Date:
December 1962
Frequency:
Quarterly[1962-]
Annual[ FORMER 1960-1961]
Language:
English
Physical Description:
v. : ill. ; 22-28 cm.

Subjects

Subjects / Keywords:
Chemical engineering -- Study and teaching -- Periodicals ( lcsh )

Notes

Citation/Reference:
Chemical abstracts
Additional Physical Form:
Also issued online.
Dates or Sequential Designation:
1960-June 1964 ; v. 1, no. 1 (Oct. 1965)-
Numbering Peculiarities:
Publication suspended briefly: issue designated v. 1, no. 4 (June 1966) published Nov. 1967.
General Note:
Title from cover.
General Note:
Place of publication varies: Rochester, N.Y., 1965-1967; Gainesville, Fla., 1968-

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
01151209 ( OCLC )
70013732 ( LCCN )
0009-2479 ( ISSN )
Classification:
TP165 .C18 ( lcc )
660/.2/071 ( ddc )

UFDC Membership

Aggregations:
Chemical Engineering Documents

Downloads

This item has the following downloads:


Full Text
'I


CHEMICAL

ENGINEERING

EDUCATION


CHEMICAL ENGINEERING DIVISION
THE AMERICAN SOCIETY FOR ENGINEERING EDUCATION
December 1962


I











REINHOLD
CHEMICAL ENGINEERING SERIES
Consulting Editor, PROFESSOR CHARLES R. WILKE,
University of California at Berkeley.
COMPUTATION OF MULTISTAGE SEPARATION PROCESSES
by DONALD N. HANSON and GRAHAM F. SOMERVILLE, both of the University of
California at Berkeley, and JOHN H. DUFFIN, U.S. Naval Postgraduate School,
Monterey, California. The first book to present the mathematics of stagewise
separation processes in the form of Fortran computer programs used to solve
separation problems in vapor-liquid processes and liquid-liquid extraction. It
will prove invaluable as a text for advanced courses in separation operations.

FLOW OF FLUIDS THROUGH POROUS MATERIALS
by R. E. COLLINS, University of Houston. A unified treatment of all aspects of
the flow of fluids through porous materials. This book is valuable to petroleum
engineers, chemical engineers, civil engineers, and soil scientists. 1961. 280 pages.
$12.50
AN INTRODUCTION TO CHEMICAL ENGINEERING
by CHARLES E. LITTLEJOHN and GEORGE F. MEENAGHAN, Clemson College,
Clemson, S.C. This book emphasizes the fundamentals upon which chemical
engineering theory is based. It contains a wealth of material on the professional
aspects of the field unavailable in other standard texts. 1959. 288 pages. $6.50
FLUIDIZATION AND FLUID-PARTICLE SYSTEMS
by FREDERICK A. ZENZ and DONALD F. OTHMER, both of Polytechnic Institute
of Brooklyn. This comprehensive work provides a wealth of data on fluid-
particle operations answering problems common to process industries. 1960.
523 pages. $15.00
DISTILLATION: PRINCIPLES AND DESIGN PROCEDURES
by ROBERT J. HENGSTEBECK, American Oil Company. Here is all the information
needed to design any distillation column for which vapor-liquid equilibrium
data is available or can be estimated. New material is presented on methods
for calculating the "splits" of the "non-distributed" components in multi-
component distillations, and for minimizing trial calculations for flash vaporiza-
tions. 1961. 380 pages. $11.50
FILTRATION
by GEORGE D. DICKEY, P.E., Consultant. A modern account of solid-liquid.
separation in wet processes: water, industrial products and wastes. It offers a
comprehensive study of filtering, including a summary of mathematical theories
and formulas and a short history of filtration development by gravity, vacuum,
pressure, and centrifugal force. 1961. 364 pages. $12.00
Two Other Outstanding Chemical Engineering Books
RIEGEL'S INDUSTRIAL CHEMISTRY, New Sixth Edition
Edited by JAMES 0. KENT, West Virginia University, with the support of a
large number of collaborators. From materials handling to product application,
this book offers a handy cross-section presentation of current practices in the
major chemical and process industries. A large number of collaborators, all
recognized experts in their fields, contribute to make this one of the most
authoritative works of its kind. 1962. Approximately 950 pages.


HANDBOOK OF VECTOR AND POLYADIC ANALYSIS
by THOMAS B. DREW, Columbia University. A probing treatment of the vector
and tensor concepts necessary in fluid dynamics, diffusion theory, electro-
magnetic theory, and heat transmission. It is invaluable as a text for engineering
students at the advanced undergraduate and graduate levels. 1961. 112 pages.
$5.50
REINHOLD BOOK DIVISION
430 Park Avenue/New York 22, New York



















CHEMICAL ENGINEERING EDUCATION


December 1962

Quarterly Journal
published by the

Chemical Engineering Division
American Society for Engineering Education

CONTENTS

Should We Abandon Chemical Technology
by Charles E. Dryden 1

New Design Methods in Chemical Engineering--
The Synthesis of Control Systems,
by Joel 0. Hougen

Instrumentation in Design,
by Kenneth A. Otto 14

Design Data and the Role of the Pilot Plant,
by W. L. Larcamp 19

Industrial Design Optimization,
by Edward P. Bartkus 23

Programmed Learning in Chemical Engineering Education,
by L. Bryce Anderme 42



Chemical Engineering Division
American Society for Engineering Education

Officers 1962-1963

Max S. Peters (Colorado) Chairman
Joseph J. Martin (Michigan) Vice Chairman
John B. West (Oklahoma State) Secretary-Treasurer
M. H. Chetrick (Louisville) General Council

CHEMICAL ENGINEERING EDUCATION, Journal of the Chemical
Engineering Division, American Society for Engineering
Education. Albert H. Cooper, Editor
Published Quarterly, in March, June. September, December
Publication Office: University of Connectiout
P.O. Box 44, Storrs, Connectieut
Subscription price, $2.00 per year.













SHOULD WE ABANDON CHEMICAL TECHNOLOGY?

Charles E. Dryden,
Professor of Chemical Engineering
The Ohio State University


Several years ago, an article appeared in CHEMICAL ENGINEERING EDUCATION by
Professor L. B. Anderson (1). Its title was "Is Unit Operations a Dirty Word?".
His conclusions were that modernized concepts of unit operations would indeed be
around for p long time. In the light of today's emphasis by many educators on
engineering science,it would seem even more appropriate to ask the same question
about technology, in particular, chemical technology.

Technology is defined herein as applied science. The prefix "chemical"
simply denotes the application of scientific principles by engineers to solving
the problems of the chemical industries. The net result is chemical engineering
synthesis--the putting together of the many facets of science and engineering
principles to guarantee performance in keeping the chemical industries in the
forefront to meet human needs. This is chemical technology in its broadest
sense. It encompasses research, development, design and systems analysis, manu-
facture, marketing and sales. All of the basic fundamentals learned in a modern
chemical engineering curriculum are utilized to a degree, including chemistry,
physics, mathematics, thermodynamics, kinetics, unit operations, and economics.
In its narrowest sense, chemical technology can be archaically described as a
highly descriptive tale of the birth, life, and often death, of many a chemical
industry. It is obvious that there is a vast difference between these two ex-
tremes as to what constitutes chemical technology and therefore how it should be
taught, if at all, to chemical engineering students.

If we accept chemical technology in the broadest sense, there should be no
doubt in the educator's mind that this type of material must be taught to chem-
ical engineers. The mechanics of teaching chemical technology have been debated
for years. An earlier commentary on this subject has been given by Koffolt (2)
in 1938 and by Withrow (5) as far back as 1911. Since chemical technology rep-
resents a synthesis concept involving a number of basic principles, we at Ohio
State have long felt that it is logical to place an integrated chemical technol-
ogy sequence in the senior year. This is supported by the recommendations of
the Grinter report, p 14, issued in 1955 by an ASEE Committee on Evaluation of
Engineering Education (2).

One of the chief questions is whether the student should vicariously exper-
ience this vast realm of chemical technology by studying in depth the patterns
of numerous chemical industries or whether he should personally wrestle with
several realistic and hopefully new problems of the chemical industries and
arrive at his own solution. While there is less room for argument among chemical
engineering teachers on this latter method, the study of industries is certainly
controversial. It is ourt opinion that both methods should be used and integrat-
ed in such a manner that the case history--vicarious experience material is in-
jected at the start of a technology sequence in a painless and interesting
manner. This combination and timing works to good advantage in an over-all
year's program of study in chemical technology.

Table 1 summarizes the course sequence in chemical technology at The Ohio
State University. During the Pall Quarter, a comprehensive survey of the chem-
ical process industries (Ch.E. 761) is introduced. This controversial teaching
concept will be discussed in detail later. Coupled with the background type of
course is Chemical Engineering Economy (Ch.E. 760). The principles of economic
balances, time value of money, and profitability analyses are typical of the
subject matter taught. The laboratory work in Chemical Engineering Economy con-
sists of several comprehensive economic analysis problems,whereas in the process
study course about 25% of the laboratory time is spent in plant visits and the
balance in library research and reporting.

In the Winter Quarter, the process development course, (Ch.E. 770), is taught
on an informal basis with the students given a typical chemical process study.
The sequence includes library research, laboratory and pilot plant experimentat-
ion, prelimary process design, and economic analysis. Several methods are
used, depending on the type of problems and size of class. The students work in
groups of 3-5 on one of several related processes or as an entire group on one
problem. In the latter case, an industrial research and development group is
simulated with assignments rotated periodically throughout the quarter. This
particular method develops management and communications skills as well as tech-
nical specialization since it is impossible for each student in a large group to
follow completely the work of others in an over-all coordinated project.

Individual solution of the AIChE Student Contest Problem, Ch.E. 790, plus
lectures on use of computers in optimization studies round out the Winter
Quarter design sequence.








CHEMICAL ENGINEERING EDUCATION


TABLE 1

CHEMICAL TECHNOLOGY SEQUENCE SENIOR YEAR
CHEMICAL ENGINEERING DEPT. OHIO STATE UNIVERSITY

Quarter, Course Number Credit Lecture
and Title Hours Hrs/Wk

\. FALL QUARTER
Ch.E. 760 Chem. Engr. Economy 3 2 2
Ch.E. 761 Chem. Engr. Processes 3 2 2


B. WINTER QUARTER

Ch.E. 770 Chem. Engr. process
Development
Ch.E. 790 AIChE Student Contest
Problem and Systems Analysis

C. SPRING QUARTER

Ch.E. 772 Chem. Engr. Process
Design
Engr. Draw. 755 Plant Design
Ch.E. 791 Special Project
Problems Investigations


2 after 30
day period



1

1


12

100
da







15
E


Laboratory
Hrs/Wk


(computation)
(25% on plant
trips)


(50% experi-
mental)
hrs over 30
7y period



6

6

(0-90%
experimental)


A chemical engineering design sequence is given in the Spring Quarter. The
process design course (Ch.E. 772) starts with a new problem for the purpose of
teaching optimization methods of process design. Digital and analog computers
are used to aid in the solution of a relatively complex problem where many
basic scientific, engineering, and economic principles influence the results.
The plant design course (Engr. Draw. 755) covers the principles of plant layout
and auxiliaries design, again using another new problem.

The special projects problem (Ch.E. 791) is usually conducted as an indi-
vidual assignment to the student by one of the professors. The scope varies
widely and may run from a design project with little experimental work to the
opposite extreme. The criterion in each case is to have the student solve some
challenging problem.

In summary, the chemical technology sequence presents the students with
five major situations which must be resolved on a professional basis. In addit-
ion, they have acquired a well-rounded knowledge of the chemical industries and
a good foundation in the key area of economics. This sequence of courses would
represent about 12% of the usual 4-year curriculum in chemical engineering.
This percentage is within the proper scope as recommended by the Grinter report
(2), page 22.

I should like to proceed next to the truly controversial technological
background course as encompassed in Chemical Engineering Processes (Ch.E. 761).
This type of course has (or had) many titles such as industrial chemistry, chem-
ical technology, and chemical process industries--to name a few. The philosophy
of our present course can be illustrated best by first giving some past history.
Before a major modernization of our chemical engineering curriculum in 1959, this
course had been taught for many years as a two-course sequence of three credit
hours each in the last two quarters of the senior year. Two text-reference type
books with a total of 2074 pages were used for study. Eight inspection trips
were taken to nearby plants with reports to be prepared. Needless to say, the
course was difficult to teach, even for professors with a great deal of prior
industrial experience. The rearrangement of the technology sequence to the
present scheme shown in T.ble 1 reduced the time allowed for acquisition of
technological background to one 3-hour course at the start of the sequence.
There was some debate as to whether this reduction in hours could be best ab-
sorbed by studying only a few chemical processes in depth or by taking a broad-
brush approach. The latter was chosen principally because all of the other
chemical engineering courses were quantitative in content and the broad qualita-
tive viewpoint often needed by engineers for prospective was lacking.

A second reason for a wide look at the chemical industries was to incorp-
orate and orient the students' background and preparation in chemistry. No other
course in the chemical engineering curriculum can better accomplish this aim.
In an analogous manner, the teaching of transport processes serves to utilize
the students' formal training in mathematics and physics. The importance of


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


TABLE 2
CHEMICAL TECHNOLOGY OUTLINE SERIES

TABLE OF CONTENTS

I. ORIENTATION
A. INTRODUCTION
B. CHEMICAL AND ENGINEERING LITERATURE
C. PLANT INSPECTION TRIPS
D. CHEMICAL INDUSTRIES FACTS AND FIGURES
E. GENERAL PRINCIPLES APPLIED IN STUDYING AN INDUSTRY

1. Chemistry
2. Thermodynamics
3. Kinetics
4. Chemical Engineering Unit Operations and Unit Processes
5. Process and Mechanical Design
6. Economics
II. INORGANIC CHEMICAL INDUSTRIES
A. SULFUR AND SULFURIC ACID
B. FUEL AND INDUSTRIAL GASES
C. NITROGEN
D. WATER
E. ELECTROCHEMICAL
F. CHLOR-ALKALI
0. PHOSPHOROUS
H. NUCLEAR RAW MATERIALS
I. HIGH ENERGY CHEMICALS AND FUELS

III. NATURAL PRODUCT INDUSTRIES
A. UNIT PROCESSES AS A STUDY BASIS
B. OILS, FATS, AND WAXES
C. SUGAR AND STARCH
D. PULP AND PAPER
E. PETROLEUM
IV. SYNTHETIC ORGANIC CHEMICAL INDUSTRIES
A. PETROCHEMICALS A FRAMEWORK FOR STUDY OF ORGANIC PROCESSES
B. CHEMICALS FROM C1 ALIPHATICS
C. CHEMICALS FROM C2 ALIPHATICS
D. CHEMICALS FROM C3 ALIPHATICS
E. CHEMICALS PROM C4, C5 ALIPHATICS
P. CHEMICALS FROM AROMATICS
G. MISCELLANEOUS
V. POLYMERIZATION INDUSTRIES
A. FUNDAMENTALS
B. TECHNOLOGY

1. Thermoplastic
2. Thermosetting
3. Elastomers
4. Fibers

chemistry is pointed out by the AlChE Committee on Dynamic Objectives for Chem-
ical Engineers in their report (5) from which I quote: "In the past, the educa-
tion of chemical engineering undergraduates has been unique in that there has
been extensive training in the parent science of chemistry. Many forces are now
tending to eliminate this feature of chemical engineering, but this committee
believes implicitly that instead, it ought to be carefully preserved and enhanc-
ed." What better place for the enhancement of chemistry than in the teaching of
chemical technology?

A third reason for the over-all viewpoint was to help the students solve
their five technology problems during the senior year with some background as to
what others had done in the past. This is much the same idea as a research
chemist doing a literature survey before making final plans for his own re-
search program.

As anticipated, we were unable to find suitable texts for such a short
course which meets for a 10 minute quiz plus a 40 minute lecture twice a week
for eleven weeks. To streamline the study program, only the Important features
of each chemical process industry were to be covered, not the complete details.
A set of notes was prepared in essentially outline form to achieve this aim. The
table of contents is listed in Tpble 2. Each industry was discussed in the top-
ical manner listed next:


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


1. Physical properties of raw materials and products
2. Consumption pattern
3. Methods of production
Chemical reactions
Process description with flow sheet
Mpjor engineering problems including thermodynamics,
kinetics, process and equipment design, corrosion.
4. Economics
In general, only one or two important processes from an industry were developed
in detail. One industry was assigned for study each lecture period,

These notes were nearly completed with many sections being sent to appropri-
ate industrial companies who have graciously offered comments for revision.
With this concise and direct study approach to each industry on the list, the
student's time is not diluted in trying to get details via the more expansive
reference book approach. Even with this new format, the student cannot possibly
absorb all of the material with its implications. However, the results of obtain-
ing a broad viewpoint of the approach to problems in the chemical industries
have been gratifying in performance achieved throughout the remainder of the
technology sequence. An added advantage accrued in giving the student a better
sense of balance in interviews and a final choice of jobs.

Teaching in this style was easier and more interesting. With the knowledge
that the students had studied the fundamental points in the outline guide of a
particular industry assigned for that day, the instructor felt free to discuss
in class some new innovation or recent engineering break-through related to that
industry.

As a result of streamlining the study of chemical technology background
material, additional time is available for individual project problems. One
assignment is doing library research and reporting on two new processes. One of
these processes is often carried on into process development the following quar-
ter (Ch.E. 770). Another part of the laboratory requirements of the course is
to make three chemical plant inspection trips as a group and then write up indi-
vidual reports.

The course is further used as training in the methods of communications.
Report writing and public speaking were removed from the new curriculum in order
to substitute more mathematics and humanities. Consequently, written reports in
the Ch.E. 761 course are severely graded and oral presentations are given with
the aid of a tape recorder for speech training.

SUMMARY

The definitions and teaching of chemical technology have been presented in
a broad, yet penetrating and modern style. The learning of necessary background
material to orient the students in a chemical technology sequence has been
streamlined to make an up-to-date, fast-moving, and interesting approach to
this study of the processes and problems of the chemical industries.
When we consider that over 60% of our B.S. degree men go directly to the
chemical industries where they immediately encounter many phases of chemical
technology, why not prepare them in advance with a mature and seasoned approach
to their new problems? This is the aim of a full year of integrated coursework
in chemical technology. This is the reason we should never abandon nor even de-
crease our teaching efforts in chemical technology for the lure of engineering
science per se. If we do, we will soon give up the heritage our predecessors
have so stoutly made for us in best serving mankind through the combining of
chemistry and engineering.
REFERENCES

1. Andersen, L.B., CHEMICAL ENG. EDUCATION, Trans. of Ch.E.
Div., ASEE, p. 13 (1960).
2. Koffolt, J.H., "The Teaching of Industrial Chemistry,"
Paper Presented at L6th Annual Meeting of SPEE, College
Station, Texas, 1936.

3. Withrow, J.R., "Points of View in Teaching Industrial
Chemistry," J.A.C.S., 33 624 (1911).
4. "Report of the Committee on Evaluation of Engineering
Education," L.E. Grinter, Chairman; ASEE, Urbana, Illinois,
(Jzune 15, 1955).

5. "Dynamic Objectives for Chemical Engineering," CEP, 57
(10), 69 (1961).


Dec. 1962














NEW DESIGN METHODS IN CHEMICAL ENGINEERING--
THE SYNTHESIS OF CONTROL SYSTEMS

Joel 0. Hougen, Technologist
Engineering Department
Research and Engineering Division
Monsanto Chemical Company
St. Louis, Missouri


In recent years chemical engineers have become aware of the success which
others have achieved in the design of high performance systems. I am referring
especially to those electromechanical designs broadly referred to as servomeeh-
anifnai where special attention has been focused on dynamic behavior and the
suppression of transients arising from disturbances.

Chemical engineering faculties have responded to the recognized need and are
including in their curricula various amounts of control-oriented subject matter
and are fostering closer affiliation with their counterparts in other departments
in order to reinforce their own efforts. As a consequence more and more chem-
ical engineers, especially those with graduate training, reach industry with ex-
cellent backgrounds in education and training in the new control-oriented dis-
eiplines.

I wonder, however, if the teachers get much feedback from their efforts.
Perhaps you may question if this special training is being used, and if so to
what ends. You may like to know the nature of the problems encountered, the
techniques used to attack the problems, and the degree of success in achieving
a solution.

Most certainly the demand for chemical engineers with a control engineering
speciality is increasing. Recruitment of personnel of this caliber becomes in-
creasingly difficult, and there is healthy competition between industry and col-
leges. The ability of industry to convert the most recent theoretical develop-
ments into reality at an acceptable rate often is responsible for talented per-
sonnel returning to academic ranks.

However, some notable accomplishments have occurred. The most spectacular
is the increased use of computers for on-line, open- or closed-loop control. In
all, a total of 159 control computers have been sold to date, about one-third to
chemical, petroleum, and allied industries. (1) For such work chemical engineers
with strong theoretical inclinations combined with aptitude for model building
and programming are in demand. The objective is to design a scheme of plant man-
agement which will improve and maintain a desired optimum. To date I do not know
of a case where the basic concepts of production have been altered because of the
availability of the computer. The process design has usually been quite conven-
tional. Special measuring and control components may be added to accommodate
the computer, but usually the computer is superimposed on the process to give
over-all guidance to a more-or-less conventionally designed control system.

The major deterrent to effecting new designs lies in the lack of understand-
ing of the complex processes found in chemical engineering. Theoretical formu-
lations have on one hand been sometimes over-simplified or on the other made
ovevcomplicated -- in each instance leading to design procedures having limita-
tions in utility. One way of eliminating the above dilemma is to use experimen-
al. data, but this suffers because of the time and expense involved in acquiring
the desired information, and indeed may be even more limited in utility unless
suitable pilot plants or prototype facilities are already available. A happy
combination of both appears to be the sound approach.

For the last five years I have been heavily engrossed in experimental work,
both with small and large scale processes. The plant work has usually been
directed toward the solution of a problem on an existing process. Thus, the pro-
cess design has been already established in the large, only minor changes being
feasible. However, auxiliary systems have required redesign, these being the
systems necessary for control.

As an illustration of a typical control system synthesis for an existing
plant, I shall describe an actual case. All steps will be described, including
the experimental work and terminating with the recommended scheme for control.
An analysis of the problem will be made followed by a description of the methods
of testing and data collection, reduction, and interpretation. The reasoning
used to arrive at an acceptable control configuration will be presented which
will be supported by theoretical deductions. Finally, it will be shown where
the experimentally determined information is utilized in order to arrive at the
best design.

Proprietary considerations prohibit revelation of the details of the pro-
cess, but this need not detract from the value of the example.








CHEMICAL ENGINEERING EDUCATION


The function of the process was to remove a condensable product from a gas-
eous mixture generated in a reactor and to discharge noncondensable gases into
a common manifold for processing elsewhere. In this particular process it was
important to maintain the pressure in the reactor at a very precise value; name-
ly, within about J 1/4" of water around some established level. It was the ob-
jective of the study to design a control system for this purpose.

The problem was made especially difficult because of the uncontrolled dis-
turbances associated with the reactor. Gas evolution, and hence pressure, with-
in the reactor was highly dependent upon the energy absorbed therein and energy
supply was subject to random variations. In addition, transitory changes in the
effluent gas temperature could occur, these giving rise to pressure changes.

The gas, approximately 15,000 cu. ft./min., was conducted to a large, dis-
tributed-type condensing unit, thence to a scrubber, and was finally removed by
compressors from which it was discharged into a main serving as a manifold for
other similar units.

Since the main pressure was subject to random variations, disturbances from
this source could also enter the system. These, however, were always greatly at-
tenuated because of the virtually constant displacement characteristics of the
compressors and the relatively high pressure gain across them.

It seems invariably true that the general configuration of a control scheme
is conceived by logical reasoning, using qualitative performance data on criti-
cal processing items. However, once a scheme is selected the detailed descrip-
tions are used to compare performance of one scheme with another or with some
established criteria. Thus, mathematical formulations do not necessarily con-
tribute to the conceptions, but are necessary for verification. This case was
typical in this respect.

In this instance some basic requirements for satisfactory operation are ob-
vious:

1. For each specific mass flow rate a given pressure gradient between
reactor and compressor suction must exist in order to assure the
appropriate flow.


PROCESS
............. ORIGINAL CONTROL SYSTEM

----- REVISED CONTROL SYSTEM


Figure 1


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


2. Since it is desirable that the reactor pressure be constant, it is then
necessary that the compressor suction change in the manner required to
insure the appropriate flow through the system.
3. Some means or compensating for very sudden changes in reactor pressure
must be made.
4. To account for relatively slow changes in the system, such as fouling
in the effluent lines and gas processing apparatus, some means must be
provided to adjust the compressor suction pressure to compensate.
5. Because the gas handling system was rather distributed, an appreciable
delay time or transportation lag could be expected.

A control system existed on the process but was not entirely satisfactory,
especially when sudden interruptions in energy supply occurred. It was also
quite sensitive to partial plugging in the lines. For these reasons and others,
a detailed study was warranted. The work began with an experimental investiga-
tion.

Twelve channels of recording oscillographs were installed, primary measure-
ments being made with very sensitive and responsive transducers. Pressures and
pressure drops were sensed with Statham strain-gage transducers, valve displace-
ments with multi-turn potentiometers, temperatures by bare thermocouples and en-
ergy by a suitable combination of variables.

Figure 1 indicates the process scheme, the location of measuring elements
during the experimental work, and the original and revised control systems.

The following tests were conducted at various levels of production with the
system on open loop or manual control.

1. Dynamic tests to determine the response of reactor, compressor suction,
and other pressures to
a. change in reactor energy input,
b. change in displacement of existing compressor bypass valve,
c. change in displacement of existing series throttling valve,
d. direct injection of an inert gas into the reactor.
2. Static tests to determine
a. performance characteristics of valves,
b. sensitivity of reactor pressure to valve displacements,
c. pressures and pressure gradients at various locations in the system.
3. Extended observations to note general process behavior, influence of dis-
turbances, and performance of control system.











L7- -- O
-- -- -- I I
-t-T 4- i-- 1














--- -i --, -.,' '
i ,- --: -" r_- l I_ '-._LL-- _,,I 0


--- TIME, 25 DIVISIONS=ONE SECOND
Figure 2


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


40


20


-irrt;;-i -I--- ~ ---I ----I I----: i..:~~-------


FO 2.2 e-045d/ / -


25 30 35


40 45 50 55


VALVE DISPLACEMENT, d, % OPEN

Figure 3



All dynamic tests were conducted by the pulse method (2,3,4) and were reduced
to frequency response via a digital computing routine, delay time between records
being extracted visually prior to processing. In this method an input is caused
to vary in a pulse-like manner and the resulting outputs measured. In Figure 2
are presented records of typical results from a test wherein nitrogen was in-
jected from a pressurized vessel into the reactor in order to measure the pres-
sure response.

Figure 6 shows the corresponding frequency response information.

Important static data were also obtained, typified by Figures 3 and 4.

In view of the nature of the system, the requirements, and the information
obtained, a control scheme was formulated.

The scheme visualized relatively fast control systems at the terminals of
the process with a means of loosely connecting the two.

1. A tight control of reactor pressure by regulation of the directly in-
jected gas stream.
2. A tight control of compressor suction through bypass control.
3. Adjustment of compressor suction pressure in accordance with energy in-
put using feedforward control of the set point of the suction pressure
regulator.
4. A means of recognizing slow changes in system behavior (fouling, etc.)
by sensing reactor pressure and allowing this to adjust the set point
of the compressor suction pressure controller.

A complete block diagram representing process and control systems is shown
in Figure 5. Symbols are defined in the list of nomenclature.

The objective of the experimental work was to determine the character of
each block. Upon completing the data reduction, information was on hand enabling
a detailed design to be made.

Since both the gas injection control and terminal bypass control systems
must be fast compared with the other control functions, each was considered sep-
arately. For compressor suction control the following criteria were used -


-i-r-- i- r--i-,-i-^i-i-^j-r-,-n-i-r,--i=r_3=c=.^i


T


Dec. 1962


~-t~t~Ti-


S..... '









CHEMICAL ENGINEERING EDUCATION


-4

















Pri nGhGnn 1
o 1- G 'nvGnnHn


(1G____ _i i f
S- -
n/ J- --N-"." .- -



































For the auction control system both criteria are satisfied if
-0 1_Z 1










A a re













AsTo maintain the re-uppiressord high gain In spite of attrubenuationes were equivalent to a


usually requires the as injection compensating dynamics. Thus ifem
Pi G 0.




Gnv Gn -- *n since then





usually requires that Gn contain compensating dynamics. Thus if


Dec. 1962







CHEMCAL ENGIMEERIN EDUCATION


Ind~~icatese a ontrollermwl variable dynamics
d-e-ined by dj--,

~Indicates a control component (sensing element,
valve actuator or a fixed control function).

Indkc.e. a proce perfo ce action.


-Indicates a umd pot


Figure 5


Dec. 1962








CHEMICAL ENGINEERING EDUCATION


anv : n and Gnn nn it is well to provide in Gn the
OT (l+rns) T=+T1 )
elements of

K (+rne) ("1'rnns)
En -nn

insofar as practicable. The objective is to provide sufficient gain and to can-
cel system denominator terms with controller numerator terms thus extending the
desired frequency response consistent with realizable components and system sta-
bility. Physical limitations exist in the control components generally available
which implies compromises.
Similarly for the injection system the criteria indicate that Gi Giv Gif
should be as high as possible. Once again controller characteristics should tend
to compensate for the drooping frequency response of Giv and Gif as frequency in-
creases.

In this manner the general requirements of the individual control systems
were determined. The next step was to consider the stability of each loop.
Using elementary servo theory the gain in the vicinity of an open loop phase
angle of -180 degrees was considered. The gain should not exceed unity at -180
degrees, and some departure from this is desirable. Many rules are available to
aid in arriving at the best performance. In this manner a satisfactory value for
controller gain was established.

The two systems were next tied together by the control functions, Gw and Gp.
The first was based on the empirical relation between energy input and requireap
compressor suction pressure. The relationship was of the form

Pn -KE




0











0 -20











QEmpirical performance function determined by curve |
, -0.4 fitting using profiles corresponding to linear forms:
W -0.8 -
a -'-- 0















-0.6 .17(+ )+ + 2(+). +


- .01 0.
0.01 0.


1 1.0
FREQUENCY, RADIANS/SECOND
Figure 6


Dec. 1962


-LU










CHEMICAL ENGINEERING EDUCATION


Also, since the corrective action should not occur before the reactor pres-
sure distrubance appeared at the compressor suction, some attempt was made to
create delay time in the control loop to match that in the process. A first or-
der time lag element was used as an approximation. The feed forward element thus
became
-K E
Gw =_ _V


Finally the block shown as Ga was selected. This control component was pro-
vided to adjust the compressor suction pressure to compensate for slow changes
in system characteristics. Gap was thus required to produce an output which was
the integral of the input. The form of control function chosen for this service
was

K (1l t Ts)


which was closely approximated by conventional components.

The experimental data were used to enable selection of control components
and for arriving at the appropriate adjustments of controllers. For example,
although Figure 6 shows that the relationship between changes in injection gas
flow rate and reactor pressure is rather complex, for part of the synthesis
the approximation can be made that the performance function is first order with
a time constant of about 4 seconds. (Reciprocal of the breakpoint frequency of
about 0.25 rad/sec).This means that a valve actuator having its largest time con-
stant equal to 0.4 seconds or less will be quite satisfactory. Thus pneumatic
components will suffice. In addition, the time constant 'I in the integration
component, Gap, should be 10 seconds or more if isolation of the terminal con-
trol system was to be achieved in some measure. The integration would then not
be effective at frequencies above 0.1 radian per second.

The compressor suction pressure control required high performance components.
Moreover, positioning of a large valve to make small changes in a large flow
rate was involved. This system called for precision components and a hydraulic
system was selected. To help compensate for the valve actuator and process
some derivative action was introduced in controller, Gn. Frequently, however,
some hydraulic components inherently provide some lead action so that none need
be deliberately added. Experimental studies on the control gear is generally re-
quired to determine if these properties exist and at what frequency.

The above approach to the design of the appropriate control system is usual-
ly adequate and optimum performance can generally be attained by minor adjust-
ments during operation.

However, if further verification is warranted, the entire ensemble may be
modeled on the analog computer and the best controller functions determined by
a systematic study.

The procedure for designing control systems for existing plants can be sum-
marized as follows.

1. Study the plant to ascertain the fundamental phenomena involved and
its operating principles.
2. Design and conduct an experimental program to determine important
performance functions.
3. Based on the principles involved and bearing the data in mind, for-
mulate the best control scheme in a qualitative manner.
4. Establish control criteria and determine the more exact nature of
the particular control functions utilizing data from the experimen-
tal studies.
5. Construct the complete block diagram, or equivalent, in order to
study over-all behavior, possible interactions, stability and per-
formance under extreme conditions.
6. If warranted, extend the study using an analog model.


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


LITERATURE CITED


1. Control Engineering, Vol. 9, No. 5, PP. 79, May 1962

2. Draper, C. S., W. McKay and S. Lees, Instrument Engineering,
Vol. 2, Chapter 25, McGraw-Hill Book Company, New York,
efi York (1953)

3. Hougen, J. 0., and R. A. Walsh, "Pulse Testing Method,"
Chem. Eng. Progr., Vol. 57, No. 3, pp. 69-79 (1961)

4. Dreifke, G. E., "Effects of Input Pulse Shape and Width on
Accuracy of Dynamic System Analysis from Experimental Pulse
Data," Dissertation presented to the faculty of Washington
University for partial fulfillment of the D.Sc. Degree,
June 1961.


NOMENCLATURE


Performance Functions

G relation between compressor suction pressure and energy
(energy signal in voltage signal out)

Gn controller function
(voltage signal in voltage signal out)

Gnv bypass valve actuator
(voltage signal in displacement out)

Gnn process between recycle valve and compressor suction
(valve displacement in compressor suction pressure out)

Hn compressor suction pressure sensor
(compressor suction pressure in voltage signal out)

Gnf process between recycle valve and reactor pressure
(valve displacement in reactor pressure out)

Opn process between energy disturbance and compressor suction
pressure
(energy disturbance in compressor suction pressure out)

G process between energy disturbance and reactor pressure
p (energy disturbance in reactor pressure out)

Gap controller function
(voltage signal in voltage signal out)

Hf reactor pressure sensor
(reactor pressure in voltage signal out)

Qi controller function
(voltage signal In voltage signal out)

GS injection valve actuator and voltage to pneumatic signal
transducer
(voltage signal in valve displacement out)

Gif process between injection valve and reactor pressure
(valve displacement in reactor pressure out)

Gin process between injection valve and compressor suction
pressure
(valve displacement in compressor suction pressure out)

Gfn process between reactor pressure and compressor suction
pressure
(reactor pressure in compressor suction pressure out)
Gnf process between compressor suction pressure and reactor
pressure
(compressor suction pressure in reactor pressure out)


Dec. 1962













INSTRUMENTATION IN DESIGN

Kenneth A. Otto
THE DOW CHEMICAL COMPANY
MIDLAND, MICHIGAN



Introduction

Instrumentation and process control has been a basic ingredient in the
growth of most chemical companies. Management has long since recognized that
application of advanced instrumentation techniques was essential in insuring
strong competitive market positions. The esteemed position instrumentation
holds today in most companies has been well earned. Its future holds even
greater potential since it represents to management one of the basic keys to
meeting competition through production of new and better products, increased
plant operating efficiencies and the ability to hold closer (and more exact-
ing) product quality specifications.

Economic Considerations

Today in a typical petrochemical company instrumentation and control
represents a substantial portion of the company's effort. According to data
on instrument sales compiled by the Department of Commerce, the chemical in-
dustry accounted for 16.6% of all the instrument sales in the country. The
petroleum industry adds additionally another 12.5%. In total, the petrochem-
ical industry accounted for about 30% of all instrument sales, yet during the
same time accounted for slightly less than 10% of all new plant investment.
The Department of Commerce figures also reveal the ratio of instrument cost
to total plant cost ranges from 3 to 15% for the chemical industry.

A large and constantly growing percentage of the money spent by many
petrochemical companies goes for instrumentation. For example, The Dow
Chemical Company usually spends between $2 and $6 million annually on instru-
ments of all types. In addition to Dow's outside purchases, we spend a sub-
stantial portion of our research dollar on engineering, design, and fabri-
cation of special purpose instruments in addition to a good share of the
test and engineering laboratory work done in the company.

DuPont released figures recently that showed an impressive 10% of their
total plant investment as being directly attributable to instrumentation, a
value of well over $100 million. Truly, the chemical and petrochemical in-
dustry is more dependant upon automatic control than perhaps any other in-
dustry.

While these figures are impressive, they perhaps fail to indicate clear-
ly the true spread of instrumentation costs on an individual project basis.
A breakdown that we at Dow have found useful is shown in Figure 1 and is a
correlation of our experience of instrument cost. In this figure, the cost
(as a percentage of total direct project cost) is shown vs the total direct
project cost. This data, which separates batch and continuous processes
shows a spread of 6 8% of total cost at an 8 million dollar project level.
A spread of 8 11% is shown at a $1 million project level for the continuous
process. Batch process instrumentation costs are characteristically less,
running on the average 4% lower than for continuous processes. Translating
these figures into dollars means, for example, that for a 10 million dollar
Slant the installed instrumentation cost can be expected to run between
00,000 and $600,000.
Design Estimates

Investment figures such as these clearly emphasize the importance of
careful economic considerations in the control system proposals. In the
engineering design area, the ability to accurately estimate the costs of
various equipment configurations and complete projects is of primary impor-
tance. In many cases especially in preliminary proposal stages, the en-
gineering estimates may make or break the complete project; while in other
cases, it materially affects selling price, profitability forecasts, and sim-
ilar areas. Since instruments and controls represent a significate portion
of the total investment in process plants, estimates of such costs must be
made with increasing accuracy throughout the progress of the engineering
project. Data such as shown inFigure 1 can be quite useful in early stages
of project study.

An alternate method of showing instrument costs is as a percentage of
purchased process equipment as shown in the next figure. Once a project has
been studied in enough detail that the major process equipment has been de-
termined, this figure becomes a more realistic base for estimating instrument










CHEMICAL ENOGIEERING EDUCATION


costs. From the figure, for a continuous process we could expect instrument
costs to be between 30 to 45% at a one million dollar project level to a 15
to 20% range at a $10 million project level. Again, batch processing in-
strumentation costs run less; in this case form 15 to 20%. This seems rea-
sonable in that instrumentation for batch operation is normally not as com-
plex as that required for continuous processes. Also, production size batch
equipment usually is quite large in comparison to continuous type equipment
and, therefore, represents a larger portion of the plant costs. The data
shown in this figure, essentially correlating past experience for The Dow
Chemical Company, is useful for two reasons; first, it realistically out-
lines instrument costs and, second, it becomes quite useful in the prepara-
tion of quickyy" estimates and in quick comparison of alternate process
layouts.

Once a project has been studied long enough to determine the various
control systems required, a preliminary cost estimate based on the installed
cost of the necessary instrumentation can be prepared fairly quickly.

Installed costs at Dow include such items as labor, painting, wiring,
piping, and similar items. Currently at Dow our installed costs for instru-
ments run between 160 and 170% of purchased instrument costs. This factor,
of course, varies from year to year and is dependent on local labor conditions
and the like.
In our detailed cost estimates, we additionally try to include all costs
of an instrumentation system such as piping requirements, power re-quire-
ments, air requirements, panel and floor space requirements and the like.
For, example, if we install a control valve, we will include in the instru-
mentation cost, the cost of the conduit run, the by-pass valve and piping
and the air or electrical requirements. These so called "extra" costs often
represent a sizable part of total. Many companies include such costs in elect-
rital' or piping areas and for this reason their instrumentation costs
often appear much lower than ours. By contrast, however, our piping and elect-
rical estimates may appear much lower than theirs.

While these figures give an excellent picture of where we have laon.
t4ey certainly do not indicate where we are going in instrument costs.


INSTRUMENTATION COST VS.
TOTAL DIRECT PROJECT COST
THE DOW CHEMICAL CO.- MIDLAND, MICH.
F I,


4 E 6
TOTAL DIRECT COST. MILLION, OF )CLLAR3


Dec. 1962








CHEMICAL ENGINEERING EDUCATION


INSTRUMENTATION COST AS
PERCENTAGE OF PURCHASED EQUIPMENT VS
TOTAL DIRECT PROJECT COST
THE DOW CHEMICAL CO.-MIDLAND, MICH.
60 I- r-
INSTRUMENTATION, % OF
TOTAL DIRECT COST

50




40 CONTINUOUS PROCESS PLANTS. ..




30




20


1 BATCH PROCESS PLANTS





0
0 I 2 3 4 5 6 7 8
TOTAL DIRECT COST, MILLIONS OF DOLLARS


In general, we can expect a continued rise in instrumentation costs
with more and more of the project dollar being spent in the control area.
For example:
A) For the past five years or so the purchase price of new instru-
ments has been increasing at a faster rate than for other equipment. This
price index rise has been in the neighborhood of 7 10% per year for in-
struments as compared to the Marshall Stevens equipment index rise of 4 7%
per year. While the instrument price rise is today tending to level off,
the effects of this difference will be felt for some time in increasing
instrumentational costs relative to equipment costs.

B) Much of the present instrument cost data, at least for The Dow
Chemical Company, reflects a minimum of analytical type instrumentation in
the initial design. Usually the analytical instruments (especially the more
sophisticated types) were installed after the plant had been in operation for
a time. More and more we find today that many of the analytical instruments
are being specified during the initial design of the project. As such, you
can expect their influence to add somewhere between 2 to 20% to cost figures
in Figure 2.
C) We are starting to reach the point of diminishing returns in the
sole use of more instrumentation. We are starting to see where something
a little better than more single element or single loop controllers will
be required. Today and more so in the future, we will see not only more in-
struments per plant but more intercoupling of instruments per plant. Ii
essence, we are slowly entering the area of more sophisticated control with
both analog and digital control schemes playing an important role. I do
not believe there is a one of us that does not feel these schemes will result
in increased instrumentation costs.

As way of example, if one takes a $20 million Styrene plant and adds
around a $300,000 investment in computer control, the total investment in
instrumentation goes from around $800,000 to $1,100,000 a sharp jump of
37%, yet with other equipment costs remaining substantially the same.


Dec. 1962










CHEMICAL ENGINEERING EDUCATION


/present Development

Today in our industrial processes, where plant measurements are of vital
concern to the economics of operation, engineers are making great strides to-
ward accuracy, reliability, and versatility of measurement. We are no longer
completely dependent on the old standbys of temperature, flow, level, and
pressure. Analytical measurements of composition and product characteristics,
rather than its environmental condition, are increasingly finding their way
into the initial plant design. We are seeing rapid improvements in measure-
ment transmission, reliability, sensitivity, and sensor accuracy and stability.
Important progress is being made in computation of significate information
not directly available from individual primary sensing elements and multiloop
performance control, accomplished by computers, is increasingly necessary.

Since most chemical processing is so dependent on proper measurement
and control, let us take a quick look at our present design trends. Measure-
ment and control of temperature, pressure, level and flow are without a doubt
still the workhorses of chemical process control. With a gradual swing to
electronics, and with refinements both in the elements themselves and in ap-
plication know-how that fully recognizes their capabilities, steady improve-
ment in these key measurements continue. Magnetic and turbine flowmeters,
capacitance, ultrasonic and radiation type level control devices are but a
few examples of important improvements in these areas.

In the electronic controls, while there is an improving acceptance in
our plants, we are still plagued by the many different signal transmission
levels. We are forced somewhat to buy a complete system from one manufacturer
rather than picking and choosing from several manufacturers as is common
practice in pneumatic. Of course, the lack of an inexpensive electronic valve
actuator is at present a severe handicap in wider use of electronics.

Increased consideration is continually being given to the value of con-
tinuous analysis equipment during the design stages of new plant construction.
Where formerly such equipment was omitted in the initial design due to delays
and lack of confidence in reliability, performance, and the like, it is now
installed as an integral part of the usual instrumentation. This is not to
imply that analyzers are as common as pressure guages for example, but it is
routine to inquire into the possibilities presented by their use in design
of new plants. Further experience with analyzers and increasing dependance
on them is largely responsible for this change along with improved equip-
ment techniques and sample handling techniques. Vapor phase chromatography
is the outstanding development of recent years in the analytical measurement
field, and is well established in many applications. While still used primar-
ily for measurement today, its demonstrated reliability is leading to in-
creased use in closed-loop control.

Although spectacular developments in VPC have tended to overshadow those
on other fields there has been considerable activity and improvement in
infra-red, in micro-wave and mass spectroscopy, and in related areas. In-
creased usage and application of these wide and diverse analytical techniques
will add measurably to the instrument engineers design capability.

Dynamic Considerations

I 'd like to switch now to another facet of instrumentation and control
that I feel is becoming more and more important in our engineering design.
For several years now, the instrumentation and control engineers have been
hammering away at the importance of process dynamics in the understanding
and application of control systems. Today a good many engineering groups in
the chemical industry have fairly well equipped (but poorly staffed) analog
simulation facilities. This effort to me symbolizes and characterizes one of
the most important advances in the instrumentation and control effort. For
the first time the control engineer has the tools and techniques with which
to intelligently compare various control schemes, to evaluate the performance
and interactions of such control, and to, at long last, come up with ration-
al justification and tangible economic benefits of instrumentation schemes.
Dozens of proposed control and design configurations can be quickly and easily
evaluated; effects of different startup and shutdown procedures can be de-
termined; operational procedures can be studied; operators and plant personnel
trained; emergency procedures can be worked out; all made with the assurance
which comes only from the intimate knowledge of both the dynamic and the
steady-state operation of the process.

As a result of active use of analog simulation techniques, we find the
so called control engineers entering into more and more of the actual pro-
cess design. In many cases, they are no longer satisfied with the process
as envisioned by the chemical engineer, but may find that to properly con-
trol the process radical changes in design or equipment sizing must be made.


Dec. 1962










CHEMICAL ENGINEERING EDUCATION


By analyzing the proposed process and its control as a system, rather than as
separate entities with dynamic as well as steady-state considerations in-
cluded the control and instrumentation engineers are able to make signifi-
cant contributions not only to the control of the process but to the entire
process aeasgn.

For example, in an exothermic reaction step, if the reaction temper-
ature starts to rise, the resulting ability of the control system to prevent
runaway conditions depends on the relative transient speed of the heat gen-
eration mechanism and the speed with which cooling can be supplied to the
system. A partial listing of typical determining steady-state and dynamic
parameters is:

Reaction Mechanism
Degree of Mixing
Instantaneous Conversion
Instantaneous Volume
Instantaneous Thruput
Instantaneous Cooling Fluid Flow
Rate of Heat Transfer
Heat Transfer Surface
Controllers; Type and Operation
Valve Type and Speed
Detector Sensitivity
Detector Lags
Heat of Reaction
In the design of a typical process, seldom, if ever, is the dynamic
interaction of all these parameters investigated even though they in-
fluence greatly the control of the plant, the operating capability of the
plant, the cost of the plant and the choice of design. In the past, it has
not been necessary to include such information because it was usually easier
and perhaps cheaper to overdesign the process to avoid problems. In this
example, an oversized cooler, extra surface areas, diluents and other exped-
ients were satisfactory. Such procedures for insuring controllability, how-
ever, are becoming less and less attractive to the chemical industry as our
competitive race tightens.

The sum total of this effort besides resulting in better designed and
better controlled plants has been to bring the process engineers and the
instrumentation and control engineers closer together. Today you will find
many knowledgeable process engineers that can discuss at will such concepts as
Bode plots, phase plane plots, and the like; while on the other side of the
fence, many more control engineers can now at least hold their own in such
fields as heat transfer, fluid flow, reactor design and similar areas. This
cross-fertilization of talents has and will continue to result in immeasur-
able benefits.

Summary

For the past few minutes we have taken a quick look at only a few of the
areas of importance in instrumentation and process control. There are others
of vital concern that have not been touched on. The steady but rapid growth
of engineering technology has brought to industry many new methods, concepts,
and tools for instrumentation and process control. The further application
of this technology to our day-to-day engineering design problems will con-
tinue to bring about the improvements in productivity and cost reduction so
prevalent in past applications of instrumentation concepts.

Costs of instrumentation, I feel, are rapidly approaching the point where
management can no longer sit idly back and consider instrumentation a nec-
essary burden of being in the chemical business. Economic justifications, so
long a part of the chemical engineers way of life, are becoming more and more
important to control engineering concepts. With the proper motivation,
tools, concepts, and hardware at his disposal however, I think most in-
strument engineers can rise to the challenge.


Dec. 1962













DESIGN DATA AND THE ROLE OF THE PILOT PLANT
W.L. Larcamp

Engineering Department

Union Carbide Chemicals Company

South Charleston, West Virginia


Engineering designs can be divided into two broad categories: (1) de-
signs which will be mass-produced, and (2) designs which will not be mass-
produced. Examples of the former type of design include airplanes, automo-
biles, television sets and even the proverbial mouse trap. These designs
are intended to be mass-produced to provide many identical items. The ap-
proach to this type of design is to develop plans and specifications on paper,
and then to construct a full scale model of the design prior to any manufactur-
ing operations. This "sample" of the design is then subjected to many perfor-
mance tests to determine if the design target specifications have been met.
Based on these performance tests, the design can be modified as required before
it is finalized and mass-production operations are begun. Although the research
and development costs for a mass-produced article can be quite high, the suc-
cess of the design itself is assured in advance of the full capital investment
in production facilities, which is an important business factor.

Examples of designs which are not intended for mass-production of the de-
sign include chemical plants, power plants and even fixed structures, sucF as
office buildings and bridges. This type of design can be called custom design
and is characterized by the fact that a full-scale model is not fiesiWbe the
prototype and the final design are one and the same. The design, as an entity,
must be committed long before it can be tested. From a business standpoint,
this means that the full capital investment of such projects must be made in
advance of a proven design. It is somewhat like "buying a pig in a poke". In
certain fields of engineering for example, structural engineering this sit-
uation does not pose any real engineering risks. There is always a degree of
personal suspense for any custom designer, but the architect or civil engineer
rarely has any doubts about the engineering sufficiency of his design. This
is true for several reasons:

1. The number of engineering variables is small, and the fundamental
relationships among these variables are well established.
2. The properties of the building materials are known.
3. Each custom design is so similar to existing proven designs that
past experience and engineering judgment can be fully utilized
to compensate for any lack of theory or precise data.

The situation is far different for the chemical engineer engaged in pro-
cess design. The number of variables involved in chemical reactions and chem-
ical processing are many, the properties of the materials handled (particularly
in mixtures) are not always well known or predictable, and the science of chem-
ical engineering is still so young that many of the engineering fundamentals
are not clearly understood. The chemical process designer thus approaches a
design problem with these general technological limitations:

1. There is no practical opportunity to build a "sample" of the
design to determine the sufficiency of the design in advance
of its commitment.
2. There is frequently no assured way to "calculate" the design based
on engineering fundamentals, either because the fundamental
data are not available or because the engineering relation-
ships are not known.

In order to minimize these technological risks, the process design en-
gineer must have certain specific experimental data on which to base a pro-
cess design. To a point, ne can over-design to compensate for the unknowns
and assure the target design performance. However, liberal over-design may
result in a technical success and an economic failure, which in the chemical
industry means the designer failed. That fact should always be understood and
appreciated by TE stitudenT, the educator and the practicing engineer. Re-
search in glassware and bench-scale development studies provide considerable
information about a process, but frequently do not provide sufficient data to
allow reasonably certain extrapolation to large-scale facilities. The ultimate
in experimentation to assure target performance is to imitate the designers of
mass-produced article and build a model. However, economics dictate that this
model must be a small-scale model elifause the design will not be mass-produced.
The designers of mass-produced articles, such as airplanes, also resort to the
use of scale models in some areas to save time and reduce development costs
prior to building an expensive prototype for final tests.









CHEMICAL ENGINEERING EDUCATION


The small-scale model employed by the chemical industry is called a
pilot plant. A pilot plant is basically a miniature chemical processing unit,
although it need not be a complete assembly of the entire process. Some of
the steps of the process may be by-passed in a pilot plant design because they
are well known and can be scaled up by classical methods. The primary role of
the pilot plant is to supply data to permit the process design engineer to span
the gap between bench-scale equipment and commercial processing facilities with
reasonable assurance. The term, semi-works plant, usually denotes a complete
miniature of a large-scale chemical process, whlhc is used to actually process
chemicals from the basic raw materials to the finished product or intermediate.
Semi-works plants are nearly always operated as miniature production units
(even though useful design data are collected) and tend to approach a prototype
of a larger-scale plant.

ADVANTAGES OF PILOT PLANTS

Many benefits can be realized by utilizing a pilot plant in the develop-
ment of a design for large-scale processing facilities. The obvious advantage
is that an opportunity is provided to test a model of the process in advance
of finalizing the design, in much the same way that models of mass-produced
designs are tested. Of course, the pilot plant is a small-scale model, whereas
the prototype of a mass-produced design is a full-scale model. Nevertheless,
the pilot plant performance can serve as a useful preview of the commercial
plant performance.

A pilot plant ideally should be "designed" in essentially the same manner
as a large-scale plant is designed. The available bench-scale data should be
analyzed and used as a basis for the pilot plant design. Whenever reaction
mechanisms or other performance characteristics are unknown, they should be
postulated from the bench-scale data and theory to help predict the pilot plant
performance. This approach forces the early assumption of a mathematical model.
After the pilot plant is built and operated, actual performance can be compared
to the design performance. Any departures from the expected design performance
of a pilot plant should be examined and explained; otherwise, the pilot plant
data provides little more than empirical, pseudo-geometric scale-up factors.
Analysis of pilot plant data can thus prove useful for establishing a true
mathematical model of a process, which can be used to calculate the plant
design. If the predicted performance of the pilot plant is realized, the
original mathematical model assumed for the pilot plant design is confirmed.
If not, the mathematical model must be adjusted to reconcile the bench-scale
data and the pilot plant data (and still be consistent with theory). These
objectives cannot always be realized in practice; nevertheless, the approach
should always be attempted in order to obtain the maximum potential benefits
from a pilot plant program.

It is often difficult to obtain consistent data on process efficiencies
and losses from small, bench-scale equipment. In many cases, the overall ec-
onomics of a chemical process are very sensitive to the useful efficiency of
the raw materials, and the financial success of a project might depend on whether
the plant efficiency will be, say 90 percent or 95 percent. The pilot plant
serves to add a needed measure of precision to the material balances (and also
heat balances) estimated for a large-scale process.

The pilot plant serves to alert the designer to many potential problems
that might not be anticipated or seen in bench-scale laboratory equipment.
The operation of a pilot plant with "real equipment" over many hours will un-
cover such things as fouling of heat exchangers or other equipment, the forma-
tion of residues, and the effect of minor contaminaitbuild-up in the process.
The instrumentation of a pilot plant often closely resembles that of a large-
scale plant, and control problems can be recognized in advance of plant oper-
ation. All of these pilot plant general observations are helpful in estab-
lishing a successful final plant design.

A very important corrollary benefit of the pilot plant is in the area of
corrosion. Pilot plants can, and in most cases should, be built of the same
materials of construction being planned for the commercial plant. Whenever
possible, specimens of other feasible materials of construction should be
placed in the pilot plant equipment to determine corrosion resistance under
simulated service conditions. Laboratory corrosion tests can frequently be
misleading, as any metallurgist can attest. In some cases, it is advisable
to construct identical parts of the pilot plant from two different materials,
one believed to be a conservative material (and the most expensive) and one
believed to be just adequate for the service. The wrong materials of con-
struction can by equally as disastrous as a poor process scale-up. Conversely,
the choice of ultra-conservative materials of construction represents a type of
over-design that can erode the profitability of a process.


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


PITFALLS OF PILOT PLANTS

In order for the pilot plant to fulfill its objective of providing useful
design data for the transition between bench-scale facilities and large-scale
commercial facilities, the pilot plant should be designed at some reasonable
intermediate scale. As a bare minimum, the pilot plant should be scaled an
order of magnitude above bench-scale facilities, say a factor of 50-100. The
exact factor depends on the plant scale-up and the nature of the process. In
any event, the pilot plant is usually large enough that its erection cost alone
represents a considerable expense. In addition, the operating costs are high.
Most pilot plants are operated on a continuous basis, and operating labor must
be provided around the clock. The design, operation, and analysis of results
for a pilot plant involves several engineers and many more operating technicians
for a period of one year or more. Thus, the total costs of "pilot planting"
are quite high and considerable time is expended. These costs (both time and
money) are expected to be recovered by a more economical plant design. How-
ever, because pilot plant costs ere appreciable, there is a natural temptation
to take shortcuts and minimize these development costs. The net result can
lead to superficial pilot plants. Unless a pilot plant operation is carefully
planned ana executed to secure the maximum potential benefits to the ultimate
plant design, it can lead to routine operation and token design data. A pilot
plant should never be allowed to become a stereotyped phase of a process dev-
elopment program.

Some pilot plants are built for the development of products that are not
characterized by specific chemical properties. For example, synthetic resins
and fibers require such empirical properties as melt points, clarity, dyeabil-
ity and "hand"- properties that result from processing techniques and that are
not always related to chemical composition. These types of materials are not
always amenable to production in bench-scale equipment and a pilot plant, or
miniature plant, is required. Such a pilot plant is utilized primarily to
produce materials or establish the necessary processing techniques. When the
time comes to design large-scale processing facilities, it often comes as a
rude shock that this type of "pilot plant" has not generated fundamental de-
sign data for scale-up. Careful planning and advance engineering studies
are necessary to avoid this pitfall.

There is a tremendous financial incentive to quickly get a new product
to market or a new process onstream. In order to compress timetables, design
work on the commercial-scale facilities is sometimes started in advance of
completion of the pilot plant work. If delays in the pilot plant program oc-
cur (as frequently happen for any experimental work), the plant design may be
essentially complete and waiting for the pilot plant data for confirmation.
When this happens, there is a great tendency to "prove the tentative design"
and thus defer or by-pass the remainder of the fundamental design data.
Such expedient action, while possibly justified by circumstances, dilutes the
effectiveness of the pilot plant and eliminates the opportunity of exploring
alternate process designs that might be more economical. Thus, a pitfall to
be recognized in the early planning of pilot plants is the possibility that
insufficient time will be available for completing the program, in which case
the full potential benefits of the pilot plant program will not be realized.

Sone modern chemical processes are extremely complex and, with the pres-
ent state of chemical engineering technology, a pilot plant cannot be "de-
signed" from the usual bench-scale data. In this case, it may also be unlike-
ly that a large-scale plant can be "designed" from the pilot plant data. A
rigorous approach to the development of these complex processes might require
years of fundamental research studies in advance of pilot plant work. The
total cost of such a program might well exceed the cost of a commercial-scale
processing unit. Thus, there are some processes for which a speculative
plant design is more economical than extensive research and development.

JUDGEMENT OF NEED FOR PILOT PLANTS

A minimum amount of specific experimental data must be available for any
new process before a design engineer can prepare a realistic plant design.
Such information as the effect of pressure, temperature, and contact time on
the yield and efficiency of a chemical reaction are usually obtained during
the preliminary research and development work in bench-scale equipment.
Physical and thermodynamic properties are either available from the litera-
ture or can be determined experimentally on a small scale. This package of
data is typical of that supplied in most design courses in chemical engin-
eering curricula and does permit a basic process design to be calculated to
some dicre.. The chemical industry must choose between two alternate ap-
proaches to chemical process design:


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


1. Utilize the data available from bench-scale studies alone and
develop the plant design from chemical engineering fundamentals,
judgement and past experience.
2. Obtain additional design data from an intermediate-scale pilot
plant to better establish the factors affecting scale-up.

The former approach saves time and reduces process development costs,
but the performance of the plant design is less predictable. The latter
approach extends the time and costs of process development, but the perform-
ance of the plant design is more predictable. Neither approach, of course,
will result in a certain design. There are no hard and fast rules to judge
which approach is better, and every new process should be separately eval-
uated and analyzed to provide a basis for decision.

For some processes, an engineering evaluation will show clearly that a
pilot plant program is not mandatory for a reasonably certain process scale-
up. If the reaction mechanisms and rates are available or can be readily ob-
tained in the laboratory (as for many slow homogeneous reactions), it is us-
ually possible to calculate a process design from chemical engineering funda-
mentals. If the potential mechanical problems that might be encountered in
large equipment are judged to be minor or solvable during a plant startup,
the pilot plant can be safely eliminated. Even if equipment mechanical prob-
lems are anticipated, there is no assurance that a pilot plant program will
provide solutions. Thus, a technological judgement alone may rule out the
need for a pilot plant.

Unfortunately, many new processes today are complex and involve such
steps as polymerizations, heterogeneous reactions and catalysis areas in
which the present state of technology is limited. It is virtually impossible
to develop a mathematical model from laboratory data, and scale-up to commer-
cial facilities is difficult to predict with assurance. Based on a techno-
logical judgement, a pilot plant may appear mandatory for a satisfactory de-
sign. However, there are other factors that are pertinent in the ultimate
decision whether to build a pilot plant. The estimated economics of the over-
all project and the policy of the particular chemical company weigh heavily in
this decision. One quantitative evaluation technique is to estimate the prob-
able range of the economics of the project with and without pilot plant ef-
fort. Such economic analysis requires good erigineerTng 'Judgement, but it is
usually possible to predict the opt'.mistic and pessimistic project economics
to a confidence level of, say 90 percent or better. These estimates, to-
gether with the sales potential of the project and the financial risks a
particular chemical company is willing to assumvie, can be used as a basis for
the decision whether to build a pilot plant.


Dec. 1962












INDUSTRIAL DESIGN OPTIMIZATION


Edward P. Bartkus
Engineering Department
E.I. du Pont de Nemours and Company
Wilmington, Delaware
Introduction

Chemical companies, like other industrial concerns, are in business
primarily to earn money within accepted ethical, sociological, and legal
restraints. Thus, the acceptance and application of any new equipment or
system depends on how it will help earn money how much money. This paper,
therefore, is concerned primarily with techniques of design of a plant which
will produce a product which will earn more money for chemical companies.
Much of the content is based on major studies of the engineering function in
the Du Pont Company, and on engineering computation, operations research
and systems engineering programs in the central engineering department.
It is important to note that the output of a design project is not a
plant, but a product. The concern of management is not primarily with the
appearance of the plant, or how much the equipment costs; but, can the re-
sulting facilities turn out a saleable product which can be sold for a
profit. Design, then, must be the effort to determine the appropriate mater-
ials and equipment which will produce the given profitable commodity. Design
must be the integrating process which selects the pertinent elements from an
array of possible alternatives.

The elephantine number of options gives the design process an artistic,
imaginative slant. For, given the same scope of work or plant objectives, no
two project engineers working independently are likely to produce an identi-
cal engineering design, regardless of how formalized and standardized the
techniques, procedures and equipment may be.

Modern mathematics and computers do make possible a logical process of
selection even with a massive list of possibilities. Mathematics can general-
ize design experience, and computers can accomplish the large volume of cal-
culations implied by the mathematical structure.

New techniques like computer control, process optimization, systems en-
gineering and operations research serve to make design procedures more rig-
orous. Terms such as linear programming, dynamic optimization, nonlinear
contraints, ridge analysis and dynamic programming enter such a picture.
But, these techniques are in their early stages of development, they have a
long way to go, and not many know how to use the techniques effectively.

Actually, much of the effort is linked together by the central idea of
trying to do something in the best way there is an optimization problem.

To repeat, although scientific knowledge and approaches are used increas-
ingly throughout the design process, the fact remains that many individual
judgments must be made. These decisions include sequence of and time to pro-
duce an individual design, materials and equipment to be selected, safe-y
factors to be considered, the amount of spare parts to be provided and the
like. Such decision problems have led to this discussion on plant optimiza-
tion.

In this paper, the following are discussed on a high-spot basis:
1. The failure of some plants to make money during the first Year
of operation.
2. The definition of optimization and "best".
3. Some techniques in optimizing the design function and process
and project design.

Why Plants Fail

A surprising number of new plants are not successful. (Fig. 1) study
by Chaplin Tyler covering the first year's performance of one hundred projects
ranging from new plants to extensions of existing plants revealed the fol-
lowing:

1. About 4 out of 10 of the projects showed a loss, or achieved
less than 50% of estimated earnings.
2. About half of those projects deficient in earnings could trace
their failures to faulty market analysis.
3. One-quarter of the project failures were attributable to process
or operating difficulties.
4. The remaining quarter were due to general business conditions
or delays resulting from start-up difficulties.








Dec. 1962 CHEMICAL ENGIIEERING EDUCATION 24




25% START-UP 25%
"25 DELAYS
PROCESS /GENERAL
DIFFICULTIES B NESA
CONDITIONS



POOR
MARKET
ANALYSIS








Figure 1





Thus technical deficiencies caused one-quarter to nearly one-half of the
financial failures. One can only conclude that a better designed facility
would likely have minimized this percentage.

Idea Inception To Commercialization

In order to discuss which techniques of optimization are useful, agree-
ment on the sequence of steps in a product venture is a prime requisite.
There are listed in Figure 2.

It is important to notice that continuing economic evaluation or ven-
ture analysis is essential to maximize the probability of success of the
project.

What Is Optimization?

In industrial design optimization, the objective is a formal, reproduc-
ible way of evaluating alternative parameters to produce some "beat" design
in terms of a predetermined criterion.

What is this criterion? Du Pont has an economic objective of a suitable
return on total investment. Others use criteria, such as five-year pay-out
period, net cash position after 10 years, etc. But once a venture has
reached the stage where a plant is being specified, what kind of plant is
wanted?

In specifying the achievement of the result "best" for it, each part of
an organization considers that his sole responsibility is to make his own
group appear as though it has been well-managed, and asks for a plant to
suit his needs. (Figure 3).

Research and process development ask for the purest product, and a
well-instrumented plant that will give lots of data, and a plant that can be
used for full-scale experiments a complex plant.

Manufacturing demands a simple design, a 100% dependable, low-maintenance
minimum instrument, minimum unit cost plant, with flexible and 130-150% latent
capacity, but which will produce one package of one grade of product so that
inventory is kept low.

Sales and distribution want a flexible plant to produce any package
size of any grade of product on demand and a high inventory of all product
packages and varieties to satisfy customers' desires and minimize loss of
sales.

Management may want a plant which gives maximum profit but which will
eliminate or minimize the risk, or at least minimize the regret. On top of
all this, the plant must be absolutely safe.

These criteria of "best" are almost mutually exclusive. Achieving each
one of these simultaneously is impossible and impracticable.









CHEMICAL ENGINEERING EDUCATION



THE STEPS FROM IDEA TO PROFIT

I IDEA CONCEPTION
PRODUCT SCOUTING RESEARCH
I
-PRELIM, VENTURE ANALYSIS
I


I I I I
PROCESS REL PROOUCT BRE MARKET DEV. VENTURE ANAL
SI CH SCALE
BENCH SCALE


Figure 2


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


MAX. PROFIT vs. MIN. UNIT COST
SOME "BEST" CRITERIA FOR PLANT DESIGN

RESEARCH & I
DEVELOPMENT SALES UNIT COST
PUREST PRODUCT SPECTRUM OF PACKAGES ".
MAXIMUM FLEXIBILITY SPECTRUM OF PACKAES
MAXIMUM COMPLEXITY HIGH INVENTORY
FULL SCALE EXPERIMENTS
MANUFACTURING MANAGEMENT -
SIMPLE DESIGN MAXIMUM PROFIT C
100. OPERATION MINIMUM RISK
MINIMUM MAINTENANCE MINIMUM REGRET
MINIMUM UNIT COST
CAPACITY VARIATION, 50-130% MAX. I MIN
ONE-PACKAGE SIZE PROFIT! COST PROFIT
ONE GRADE OF PRODUCT
LOW INVENTORY x y
PRODUCTION ILBS.---

Figure 3 Figure 4




Let's illustrate with a simple example. Plant operations often strive
for minimum unit cost of product. Engineers design the process and equip-
ment accordingly. The plant, however, should have been designed for maxi-
mum profit. Figure 4 illustrates the difference between minimum cost and max-
imum profit. Dollars (profits, costs) are plotted against units of output
in pounds. As output is increased from zero, the average unit cost of product
is high at low outputs, falls as output is increased, reaches a minimum and
then often turns upward.

The reason for this is that at low outputs the fixed charges must be
apportioned over a few units of product, resulting in a high unit cost. At
medium outputs (between the dotted lines), the same fixed charges are divided
by a substantially higher output, whereas variable unit costs are still
reasonable.

As outputs are pushed higher, perhaps beyond originally planned capaci-
ty, operating costs start climbing rapidly. Overtime and night differential
go up, higher maintenance is required because operation is higher than at
rated capacity, unit sales go up but gross sales dollars may go down because
of lower prices from over supply or because of competition.
The result is that the profit curve may go through a maximum before
minimum plant operating unit cost is achieved.

The above example illustrates the necessity of venture strategy planned
according to predetermined criteria.

In a new venture, the best plant or facility is one of a set of alter-
native design schemes, anyone of which could be developed within the physi-
cal, economic and social constraints prescribed by management. This plant
must be fully described by the set of actual consequences which differs least
from the desired consequences, and must answer many of the following ques-
tions:

What is the plant to do in terms of (Figure 5):

A. Product (grades, packages, color).
B. Performance and reliability (size, duration, efficiency,
appearance, maintenance, safety).
C. Cost and Profits (absolute, relative, or competitive situation.)
D. Time (when product is wanted, need for capacity increase.)
E. Other constraints (risk or regret).

Introduction To Optimization Techniques

From the point of view of a theoretical mathematician, the optimization
of plant design should present no difficulties.

However, mathematical and computer approaches in industrial design optimiza-
tion are still limited. The costs of mathematical analysis, systems engineer-
ing, special tests, computation and simulation, must be included in the cost
of engineering. Many of the decisions in design, of course, do not as yet
lend themselves to mathematical quantification and must be judgment decisions.,


nBe. q1962









CHEMICAL ENGINEERING EDUCATION


MINIMUM DESIGN CRITERIA COSTS OF OPTIMIZATION vs. BENEFITS


PRODUCT

PERFORMANCE AND RELIABILITY
COST AND PROFITS A
TIME

OTHER CONSTRAINTS



Figure 5 COSTS-


Figure 6


As shown in Figure 6, the benefits may accelerate (Curve A) as the de-
gree of and therefore the cost of optimization increases, ot the benefits
may taper off gradually (Curve B). (We don't know yet the answer to this
question.)

Even if one maintained some degree of optimism, the cost of optimizatior
will still be high. To decide whether such sums are worth spending, it is
necessary to examine the possible benefits of the proposed installation to
decide whether the expenditures are justified. Once this is done, the rest
is straight forward. The proposal merely competes with other projects for
the available investment money. If the pay-out is good, the more rigorous
approach will probably be authorized. If others are better, it may be delay(
or not done. One suggestion frequently advanced is a trial installation
under a research basis to find out what benefits result. Fortunately, many
ideas compete for the research and venture dollar, so the program with the
payoff is authorized. Further, rigorous optimization may cause project time
delay and as such may be rejected.

It is worthwhile to illustrate at this time the complexity of a chemi-
cal plant project (Figure 7). Without trying to discuss the chart in any
detail, one can see that over-all plant optimization is a formidable problem
whether or not mathematics and computers can ease some of the complexities.(

In view of such complexity in industrial design optimization, techniques
within three areas dealing with a project appear to justify systematization
and discussion.

1. Engineering function, or how the project is managed.
2. Process design, which deals with selection of the equipment
to produce the product.
3. Project or detailed design, which specified the mechanical
aspects of the equipment.

Some of the techniques will be covered briefly, and others will be
described in a little more depth, primarily for purposes of emphasis and to
illustrate the present state of development. No procedure is explored
enough to make possible immediate application. This, too, has its purpose -
in most instances, the techniques have a long way to go to become standard.

Optimizng The Design Function

Optimizing process design and project design carries with it the need
to optimize the design or engineering function how best to manage the
project. Good management or project engineering divides itself into two
parts:

(A) Using the proper, economically justifiable tools which
get the job done.
(B) Allocation of resources to match the cost and timing
of the project as established by management.

Several techniques in the first group are:

1. New methods of data and information storage and retrieval.
2. Scale models.
3. Standards.
4. Analog simulation.

And the last requires application of modern planning and scheduling
techniques.


Dec. 1962









Dec. 1962


CHEMCAL ENGINEERING EDUCATION



The Project Engineer


Figure 7 Project organization.


RASE, H. F., AND BARROW, M
PROJECT EGlE-RIN OF PRoCESs PLATS
J. WILEY SoNs (1957)










CHEMICAL ENGINEERING EDUCATION


PROJECT COST VS. TIMING



CRASH



I I EXTENDED



0






T, Ta TE
DURATION OF PROJECT --


Figure 8


Tools Used On The Job

1. Information Storage and Retrieval
Computers are particularly useful for this activity. A com-
puter library of physical, chemical, and thermodynamic data will
pay off fully in manhours and savings. In cataloging of similar
drawings by number and description, one can effect substantial draft-
ing savings where repeat design is done. Such a course of action
requires only managerial vigor.

2. Scale Models
Use of models to depict the process arrangement, piping,
electrical,instrumentation, and heating and ventilation, has re-
sulted in many advantages. Such models eliminate the need for
preparation of piping arrangement drawings. They permit a more
thorough review by plant production and maintenance people as de-
sign progresses. It is much easier to spot and correct many po-
tential safety and fire hazards and interference.

Models permit better planning by the construction forces for
sequencing of installation of all components, and, of course, are
very helpful in training of plant operators. Models have provided
unexpected bonuses in foreign work in that they assist greatly in
overcoming the language barrier.

3. Standards
Use of Standards is an excellent cost cutter, since they repre-
sent the best known practices. In turn, maintenance costs are re-
duced since plant maintenance forces help develop the Standards as
well as use them for maintenance of the plant. It has been found
that economies gained from common agreement on certain standard
varieties more than offset costs of "bending" of design to meet
availablematerials.

41. Analog Simulation
This technique has proven to be a valuable tool in analyzing
expected operation of a plant prior to start-up. Although it takes
considerable manpower, time, and money to simulate a process, it is
often justified for portions of the more complicated continuous pro-
cesses in which there are many recycle streams and many variables.
Simulation gives a good check on the instruments control system and
permits going through start-up, and steady state operation. It is
quite easy to introduce upsets of all kinds and to check the ability
of the control systems to restore operations to normal. Such trials
make possible the discovery of shortcomings which would otherwise
not be found until the new plant has been started.
Justification for the analog simulation then comes from reduced
time for startup of the new plant to obtain the desired capacity.


Dec. 1962









CHEMCAL ENGINERING EDUCATION


CRITICAL PATH METHOD


















0 1 2 3 4 5 6 7 8 9 10 11
TIME

Figure 9


Allocation of Resources
Project management has to balance resources against demands. Experience
has shown that there is a minimum total cost for the correct project time -
project design/construction relationship, as shown in Figure 8.

Total cost goes up if the project timing is extended, and total cost
increases if management calls for an accelerated plant start-up. Project
management also has to balance the time and cost required for process and
equipment optimization against the benefits.

One new tool increasingly being used by design organizations is the
Critical Path Method (CPM). The PERT system of the Navy is a similar tech-
nique. Critical path planning and scheduling (Figure 9) is a project man-
agement tool that leads to the best combination of timing, cost and equipment
optimization in projects. The technique is based on an arrow diagram-which
records the logic of the problems in a graphical manner, and states the-pro-
ject activities in the sequence in which they are to be performed. The ar-
row diagram provides the planner with the following benefits:

1. It provides a disciplined basis for planning the project.
2. It provides a clear picture of the project scope that can-be
easily read and understood.
3. It provides a vehicle for evaluating ultimate strategies and
objectives.
4. It pinpoints responsibilities of related departments by showing
the interconnections among the jobs.

Basically, the method lets the manager know precisely which steps in
the project are critical to completing the projects on time. He then "bird
dogs." The noncritical steps have some leeway or "float time." So, if they
are late in being finished, the completion time of the entire project is un-
affected. There is no "float time" on critical steps. Obviously, then,
these steps are on the critical path. The critical path method is different
from other methods, such as a bar chart, because it separates the planning
and scheduling.

CPM not only permits optimizing manpower among several projects, but
presents a road-map which matches rigorous approaches versus time and bene-
fits.


PROCESS DESIGN OPTIMIZATION

How Bi_.A Plant .- hB Which Process?

For the process industries, capital investment costs have increased per-
haps six-fold over a 30-year period. Such increases have made mandatory care-
ful look-see at just how much capacity is actually designed into a plant spec-
ified for a nominal figure.


Dec. 1962








CHEMICAL ENGINEERING EDUCATION


MARGINAL ANALYSIS


INVESTMENT vs. RETURN ,














$ INVESTMENT ( CAPACITY) $ TOTAL INVESTMENT----


Figure 10 Fieure 11
More often than not, sizing is based on the oversimplification that cap-
acity should be what the sales organization can expect to sell. (Figure 10.)
Underoapacity minimizes initial investment, but also limits capacity to meet
foreseeable market demands. Hidden overcapacity wastes investment dollars,
dilutes profits realized on the marketable amount of product, but permits at
some future date meeting expanding markets with relatively little additional
investment. It is important that cost of over-design(primarily wasted capac-
ity) should balance that of under.-des.gn(lost potential profit). An optimum
capacity plant, however, minimizes investment per unit of capacity, and yields
maximum profit-investment ratio.

To discriminate from among a number of alternative proposed plant capac-
ities or plant processes, such criteria as return-on-investment and five-year
payout are used. However, in a considerable number of instances, additional
yardsticks for comparing one investment opportunity against another are be-
coming necessary to make the selection process more effective. It is sug-
gested that a several step screening process be followed. For example, first
plot return vs. capacity (or investment). Eliminate those which do not meet
corporate statutory return levels. Subject the remaining candidates to ad-
ditional screening processes. It should be remembered that at an early stage,
where the data are broad range or inaccurate, some alternatives may be elimi-
nated. A review of past decisions in the light of new information is often
justified.
Marginal Analysis (Figure 11)
(2)
This approach offers one method of solution to this problem. The
procedure is straight forward:
1. Calculate a base case.
2. Calculate incremental investment to get each increment of
additional capacity (11, 21, 31, etc.)
3. Determine savings or return attributable to the incremental
investment. (1R, 2R, 3R, etc).
4. If the incremental return meets statutory levels, the larger
capacity is specified. (Point "Z" does not.) This approach
guarantees that the last dollar invested earns the required
money.

Cash Flow Method

Another screen could be the discounted cash flow method. In the ex-
ample (Figure 12), each of the alternatives showed the same return on both
the original and average investment. However, as can be seen, the annual
income for each case varied considerably. By discounted cash flow calcula-
tion, Case A came out ahead and appeared to be twice as good an opportunity
(24% vs. 12%) as Case C.
There are a number of similar screens. However, at least one more
screening bears mention at this time. This is the use of decision rules in-
volving consideration of risk, uncertainty and intangibles.
Decision Rules
In these, the expected profit under each contingency can be put into a
simple matrix form. Some examples of the possible criteria are: (Figure 13).
1. Maximin Criterion.
With a pessimistic and conservative strategy, the smallest profit
for each alternative is examined under any contingency and the


Dec. 1962










CHEMICAL ENGINEERING EDUCATION


highest value of this smallest profit selected. This criterion
maximizes the minimum profit or maximin.
2. Maximax Criterion
With an optimistic and speculative strategy, the largest profit
for each alternative is examined under any contingency, and the
alternative selected which has the highest value of the largest
profit, or maximax
3. Maxim Criterion.
If the strategy is neither completely conservative nor complete-
ly speculative, but somewhere in between, a criterion measure can
be maximized using any proportions of the maximin and maximax
criteria. For example, one might feel 70% conservative and 30%
speculative; or, one might be the reverse, 30% conservative and
70% speculative, in which case the values would be different.
4. Minimax Regret Criterion.
Under this strategy, the maximum potential losses are determined
for each alternative and contingency and that alternative is
chosen which has the smallest potential regret.

Process Optimization Individual Eqguipment:

As previously stated, cost and time of engineering the "best" design must
be balanced against the significance of the equipment being investigated, in
terms of its proportionate part of the investment, the chances of reducing
that investment, achieving more reliable operating equipment, and meeting the
project schedule.

There is a spectrum of design optimization approaches varying from su-
perficial, qualitative, to the most analytical, quantitative.

Wide Set of Alternatives

One can use charts in selection for the best among a set of alternatives.
Take a liquid-solid-separation system, for example, and particularly a filter.

The process design engineer must consider both the performance as related
to materials handling, filtrate quality, cake quality, etc., as well as econom-
ics including initial cost, installation, operating, maintenance and replace-
ment cost. Filter manufacturers do supply certain performance information
based on design, testing and experience. They can assist in developing a cost
picture. However, the design engineer is responsible for the system design.
The design engineer must accumulate sufficient information so that he can con-
sider the effect of as many parameters as indicated by the sensitivity of the
materials being processed.

Often, several types and variations of equipment can perform essentially
the same duty. To optimize in a semi-quantitative sense, charts should be
developed in which the performance characteristics of each variation can be
.compared against a common base.



RETURN ON INVESTMENT VS. CASH FLOW DECISION RULES FOR INVESTMENT SCREENING
Annual Income
40 --- --- I --- -- i ---
A 1. MAXIMIN: PESSIMISTIC, CONSERVATIVE
.. ..HIGHEST VALUE OF SMALLEST PROFIT,

2. MAXIMAX: OPTIMISTIC, SPECULATIVE
HIGHEST VALUE OF LARGEST PROFIT.

C 3. MAXIM: X% CONSERVATIVE, Y% SPECULATIVE.

4. MINIMAX REGRET: LEAST REGRET.


5 EARS 20 2Figure 13

A B C
RETURN ON ORIGINAL INVESTMENT 12 121 12%
RETURN ON AVERAGE INVESTMENT 24% 24. 24%
RETURN BY DISCOUNTED CASH FLOW METHOD 24 15.5% 13%

Figure 13
Figure 12


Dec. 1962











CBEICAL ENGINEERING EDUCATION


Dec. 1962


HO Oi IZO' TAL v CUUOI F RTEl r+ S D kT -Li LC


O1- IN D. aN 0E0, Pc ..40
SCRBON TEE.


4VAcuu D NU r TE
TYP 304ST T



25z 1 LCUU GNU PI TE

RARO N EELtEl






TCARO STE 304 T E TI 4 Issr SI ER
SCLR F ATE o STEE "E l '
CA I I ]3 e















NOTE:
I ALL VALUES BASED ON FILTER AREA ONLY-NO ACCESSORIES ozo

i-2 N EACH CASE VALUES ARE BASED ON THE MOST FREQUENTLY I
USED DESIGN MODIFICATION

I Ls a Ls 3 A 5 BA 6 9 F. IS SO So N N SOS SRI O o 1.550 BOB 260 SN i E SU* .I E 4R0S


TOTAL FILTER AREA-SQUARE FEET
Estimating Cost Chart
Figure 14


..THE TEXTILE FIBER FAMILY TREE


NATURAL FIBERS



Animl [ineral Vegetoble Synthsolic


Asbestos Cotton
Hemp
Linen
Sisol


S. EHLERS, IND. ENG. CHEM., S3 NO- 7,
P. 552 (JULY 1951). COPYRIGHT tIS1
BY AMER. 0HEM. SOC. AN0 REpRINTE
Dy PERMISSION OF Co1PYR)GHT OWNER,


Acerbate
Acrylic
Modacrylic
Nylon
Nyr I
Olefin
Polyester
Rubber
Soran
Spandex
Vinct
Vinyon


MAN MADE FIBERS'



Regenerated





Rayon Rubber Glass Azlon
Metallic




Cenec 7met for man mode dbers. by he T-xtile Fiber Prod~., Ideali.
fiction Act approved Sepembbr 2, 1958, .mad. *ftfive by Fd.r.ol Trode
Comminion March 3, 1960,


Figure 15


COPYRIGHT 1i5 By A-ERICAN

PE uICao oI O o R ... INTED









CHEMICAL EMGInEERING EDUCATION


VARIABLES IN FILTER
FABRIC CONSTRUCTION

GENERIC TEXTILE FIBER
CROSS-SECTION
CRIMP
YAM-TYPE
MONOFILAMENT PLIED
MULTIFILAMENT SINGLE
SPUN-STAPLE
YARN SIZE AND WEIGHT
TWIST AND PLIES
THREADCOUNT
WEAVE
PLAIN, TWILL, SATIN, FABRIC FINISH

Figure 16


FILTER CLOTH SPECIFICATIONS


MAXIMUM PRODUCTION RATE

MAXIMUM SERVICE LIFE

MAXIMUM GASKET ING

MAXIMUM SOLIDS RECOVERY

MINIMUM BLINDING

MAXIMUM CAKE RESISTANCE


Fi gure 17


ECONOMICS


MAINTENANCE


PROCESS
PRODUCT


PRODUCTION
Van Note, R. H., I & ECem., 53, 546 11%1)

Figure 18


This permits consideration of a sufficient number of alternatives to
result in a good or "best" design. An example is Figure 14, an estimating cost
chart, in which the relative value per square foot of basic filter types in
available sizes and materials of construction is plotted against the total fil-
ter area in square feet. (3)
Detailed Qualitative Analysis
Very often, qualitative investigation in depth is critical to selection
of the best material and equipment, a filter medium, for example.
The filter medium, organic or inorganic, may be:
a. Granules or powder.
b. Porous, perforated or sintered sheets.
c. Fibrous woven fabrics, nonwoven felted materials, mats.
If fibrous, the medium may be selected from one of a large number of
generic types (Figure 15) whose general classification include, animal, min-
eral, vegetalbe and man-made fibers. (4)
The variables in filter fabric construction (Figure 16) must be eval-
uated. (5)
Further, these parameters have to be considered in the light of the such
desirable filtering characteristics as in Figure 17.
Some characteristics may suffer in order to enhance others. And to add
to the confusion, one must consider the manufacturer of the original fiber and
the manufacturer of the filter.
Analysis in depth may mean the difference between success and failure
of the filtration system. Where time allows, lab tests, of course, will be
of help in screening.


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


PREPARE INPUT



SCAN INPUT DATA



SELECT PRELIM, C
THICKNESS


F SELECT i p


COMPUTE TILTRATI
TIME


ADJUST Ap,Li,ef
TO ACCEPTABLE
COMBINATION

t ...


Figure 19

Assigning Comparative Quantitative Values
A more rigorous approach could be the use of the "Pyramid Merit System."
The procedure is as follows:

1. Select the four critical parameters. (Figure 18.)
a. Process and Product Stoichiometry,conditions,
yield, product quality.
b. Maintenance Replacement, prevention problems,
and cost.
c. Production Process operation.
d. Economics Effect of improved product quality,
depreciation, interest, taxes, return.
2. Prepare a diagram representing the four variables. (6)
3. Establish a merit system for each of the dimensions of con-
sideration which enter into the selection of the equipment.
4. Reduce this merit rating system to numerical values, even if
only comparative.
5. Plot the numerical values so that they can be balanced one
against the other and an optimum solution obtained.

The perfect solution to an equipment unit problem would be defined by a
point equidistant from the four apexes of the diagram. However, materially,
selection is made of the specific equipment which only approaches this point.

In this type of evaluation, care must be taken to give each item its
proper weight in relation to its over-all contribution.

Computer Optimilzat ion of Equyipment

In a still more rigorous approach, where possible, parameter, physical
property and cost correlations are prepared in use of a computer tc do the
many calculations required to optimize the design. Such nr approach has
been found to be of use in selection of a drum filter. (Figure 19).

After the input data (1) have been prepared, the computer (2) scans the
data and instructions provided, to establish that the problem is not in-
determinate for lack of adequate information. (3) Unless instructed different
ly, it next selects a reasonable value for cake thickness and a pressure drop
(4) across the cake of 10 lb./sq. in. and (5) computes filtration time. If
the time is outside prescribed bounds, the computer will adjust pressure drop
until an acceptable combination (6) of pressure drop, cake thickness and time
is found. (7) During this phase of the calculation, the effect of filtrate
vapor pressure is taken into account. If, for any reason, a suitable com-
bination of variables cannot be found, the difficulty is Identified by a


Dec. 1962


OPTIMIZING SELECTION OF DRUM FILTER


DATA CONSIDER FILTRATE CALCULATE DRUM SPEED
11)J VAPOR PRESSURE (7) (13)


CALCULATE FILT, WASH CONSIDER PARALLEL
I2) & DRYING TIMES 8) FILTERS 114)


AKE ADJUST DRUM SPACE SIZE AUXILIARIES
03) ALLOTMENT 19) 1151


14 ALLOT SPACE FOR CAKE CALCULATE INVESTMENT
REMOVAL 110) COSTS 16)


CALCUIAE DRUM AREA CACULATE INCREMENTAL
S1 OPERATING COSTS (17)


CALCULATE DRUM FIND MINIMUM VALUE OF
16U DIMENSIONS 112) OP.COND. & MIN INC.OP.COST
FINAL SPECS1!8)


PRINT FINAL SPECS.Ji9)








CHEMICAL ENGINEERING EDUCATION


suitable typewriter comment and the computer stops prior to acceptance of
new data.

(8) From the selected variables, the computer next calculates the time
required for the filtration, washing and drying operations. (9) The angular
allotments for each phase are computed and checked for practicability, with
suitable adjustments being made, until a workable assignment of drum space is
achieved. (10) During these calculations, appropriate angular allotments are
made for cake removal, waste space and resubmergence, based upon accepted
practice for vacuum drum filters of various sizes. (11) The drum area is next
calculated and the drum dimensions (12) selected from tables of standard filter
sizes. (13) Drum speed is computed and compared with accepted practice. If
too high, the cake thickness is increased and the entire procedure repeated.

(14) If more area is required than can be provided in a single filter,
the smallest number of parallel filters of equal standard design will be spec-
ified. If the selection of a standard designresult sInEhe provision of more
than the needed surface, the computer will take advantage of this circum-
stance to reduce vacuum pumping costs by reducing the proposed pressure drop
to the extent permitted by the surplus area.

(15) Next, the vacuum pump and piping is sized and the investrent(.6)
required for the entire installation computed.

(17) The operating costs for the entire filtration operation are com-
puted, including the assignment of an appropriate value to the solvent loss
through the vacuum system. It should be noted that this "Incremental mill
cost" includes only those items of cost which are functions of the operating
and design variables discussed above. A full mill cost is not computed nor
should it be inferred from the reported results.

(18) Unless instructed otherwise, the computer will now proceed to ad-
just operating conditions until it has found a combination which results in a
minimum value for the incremental mill cost. This search is initiated by re-
ducing the pressure drop until the next larger standard filter will be required.
Reduction of pressure drop continues until all the area of this larger filter
is fully utilized, whereupon the incremental mill cost for this installation
is computed and compared with that for the earlier filter. This process is
continued until three filters of increasing size are observed to result in no
net mill cost improvement. (19) At this point the computer will type out the
cost of the best combination of variables located and stop. Values for the
operating and design conditions for all designs explored are displayed, to-
gether with the applicable investment and mill cost estimates, for ready
comparison between cases.

Process Flow Sheet Optimization

Suboptimization of individual equipment items certainly makes for bet-
ter design and operation of that functional unit or engineering operation.
But what effect does this suboptimization have on the entire process? If the
sequential steps have minimum interaction, other than product input and out-
put (for example, with no recirculating streams), such suboptimization con-
tributes fully to obtaining the "best" over-all process.


ACETYLENE PROCESS ALTERNATIVES



F F I F] 1 PROBLEM
DETERMINE
mov,. ,s 3 FEEDS OPTIMUM
ARC M.CTo / PROCESS
111 ''RN i* ROUTE

11 REACTORS 33



_.,[ l L w ..
3 COMPRESSION ROUTES 99




5 SEPARATION ROUTES : 495


Figure 20


Dec. 1962















(43) Pzex

(44) Poe

(45) z1'1L


(46) Ior1'


(47) Co001

(48) F Tr


CHEMICAL ENGINEERING EDUCATION



PROCESS MODEL (REACTOR OUTPUT)

24943


-" 771


71
P ex
y2,


s 120.14 i"-+i l
227.98 Y I orl
[ni3ai + zori ]
Y3 r


(49) : T.r 92.13 r rl or' 9 Corl2- C.t
227.98 1 .l120.14-

Figure 21



Should the operating conditions of some steps bear on the successful
operation of others, if there are recirculating streams, if there is inter-
action, then individual equipment optimization may adversely affect other
phases of the process.
Qualitative Optimization
To try to apply rigorous, formalized techniques in all cases would be
ridiculous. For example, consider a case (Figure 20) where some hydrocar-
bon, such as acetylene, propylene, or ethylene, is to be produced. The
problem is to select the best process from the feasible alternatives, subject
to certain constraints, such as time, competition, plant location, utilities,
etc.
Assume three kinds of feed stock. Each feed may be pyrulyzed in, say,
eleven different types of reactors. The crude stream from the reactors can
be compressed in any of perhaps three different types of equipment and pur-
ified in any of perhaps five purification trains. Without considering all
external environmental factors, there could be 495 possible process combi-
nations. In today's state of the art, rigorous approach to selection of
the best of the 495 is out of the question. Many of the combinations are
likely to be incompatible and can be so determined on a judgment basis by
experienced engineers. However, once the reasonably possible routes are
selected, increasingly sophisticated techniques can be used to select the
best process.


COST MATHEMATICAL MODEL


INVSTMiENT
DISTILLING COLUMN
PACKING COSTs 1V(H-51 D2
144


LABOR COST =
13(220) (H)O.(ID)O n F


Figure 22


EVALUATION OF ALTERNATIVE,
(CASE APPROACH)
NO. FEED REACTION PURIFICATION

1 F1 Partial Oxidation Absorption
2 F2 Partial Oxidation Distlilation
3 F3 Partial Oxidation Molecular Sleve


Pyrolysis


Distillation


Figure 23


Dec. 1962


* Cotr


PERCENT PERC
RETURN ACCUR
19 t3
10 t4




18 1(


REACTOR
rI









CHEMICAL ENGINEERING EDUCATION


Case Approach with Computers

The case method is an optimization approach. As experimental work pro-
ceeds, and the flow sheets are being developed, the flow sheets are described
mathematically, Both heat and material balances and alternative process steps
are considered. Here is an example (Figure 21) of a reactor which has eight
feeds and one discharge. The set of equations is presented as an illustra-
tion of the type of expressions required to specify a process mathematically.
Engineers do this analysis anyway, but this approach formalizes the effort.
The resulting equations are put on a digital computer and these equations
varied as to operating conditions. Alternative processing steps can also be
considered. At the present rate of development of the logic, 300 to 500 es-
sentially algebraic equations describe a flow sheet. In order to be solva-
ble on a computer, they must be appropriately ordered or arranged so that a
solution can be achieved in a reasonable amount of computer time. If just-
ified, optimization mathematics will do the selection quickly and automat-
ically.(7)

It is now necessary to introduce (Figure 22) investment and operating
costs so that a logical choice can be made from among alternative processes.
To inject cost and investment information, effective conceptual design is
essential. Iccurate fabrication cost, design, and operating information
is required. The chart typifies two of the equations required for a typical
process step. In a manner analogous to thatfor the process mathematical models,
expressions representing the cost for each step can be derived. For each step,
investment or other profit criterion can be expressed in terms of the same
variables of time, temperature and concentration. Next must be found the com-
patible values of the process operating conditions and plant costs which will
maximize the return on investment.

Once the flow sheet models and the cost models have been derived for
various alternatives, these models (Figure 23) can be manipulated via com-
puter to produce an output from which can be selected the choice processes
for further investigation. For example, from this chart, within certain
criteria for such selection, management might choose Case 1 and Crse 2 for
further investigation because of the medium high returns combined with min-
imum uncertainty about those- returns.

Or it may be that a higher degree of precision is needed to exercise a
logical choice among alternatives. It will be necessary then to iterate or
repeat the program until laboratory and cost information have been refined
to the desired degree of accuracy. Based on these factors and many others
such as timing, risk, and other economic considerations, the appropriate lev-
el of management chooses one or more courses for further development and
eventually for the commercial facilities.


VAPOR (GAS) CORRELATIONS AT "MODERATE" PRESSURES


ft NSc

PURE & MIXTURES
Npr ----MIXTURES ONLY


Figure 24


Dec. 1962










39 CHEMICAL ENGINEERING EDUCATION Dec. 1962



There are a number of more complex optimization approaches in develop-
ment by a number of companies, including "Cheops" (or Chemical Engineering
Optimization System) by Shell Development Company. The very fact one sees
so little detail in literature should be an incentive for the universities
to pioneer rigorous systems.

Project (Detailed Design) Optimization

Once the process design has been specified, many design details have to
be determined before the equipment can be procured and built. Such detailed
calculations lend themselves to rigorous methods. So, more and more Computer
programs are being developed which are intended to produce the best plant
design for an equitable cost. Two different programs illustrate this prin-
ciple.

Computer Calculation of Physical Dnta (Figure 24)

Hand computation and library search time can be reduced substantially
with an added plus in reduced errors achieved by using computers. Physical
property constants of the most commonly used liquids and gases are stored in
the computer library. The most frequently used properties are expressed by
equations. Such correlations are applicable to the widest range of materials
and independent variables and are readily adapted for calculation of mixture
properties. The flow chart in Fig. 24 is a road map for gas correlations via
computer programs. For example, the Prandtl number can be considered a main
computer routine with the thermal conductivity as the first sub-routine,
viscosity as the second sub-routine, critical constants as the third, and heat
capacity at constant volume as the fourth. With molecular structure as the
only input data, computation proceeds until the Prandtl number is obtained.



HEAT EXCHANGER DESIGN







CALCULATE NUMBER OF TUBES LOOK UP FROM TABLES
SHELL DIAMETER, SHELL THICKNESS NOZZLE WEIGHTS,
NOZZLE COSTS

IS X.RAY AND/OR
STRESS RELIEF REQUIRED ? CALCULATE TUBE WEIGHT,
TUBE COST


CALCULATE TUBE SHEET THICEESS,
CALCULATE SHELL THICKNESS AMET E EET C O T
5535 SMELT CS$T

CALCULATE X-RAY AN"O R
STRESS RELIEF COST CALCULATE NUMBER OF BAFFLES.
BAFFLE WEIGHT,
BAFFLE COST
CALCULATE SHELL WEIGHT,
SHELL COST.
INCLUDING HEAD AND FLANGES CALCULATE CHANNEL LINING COST,
SHELL LINING COST

CALCULATE CHANNEL AND
COVER WEIGHT, CHANNEL AHN CALCULATE SHELL ASSEMBLY COST,
COVER COST BUNDLE ASSEMBLY COST.
MISCELLANEOUS SHELL AND
BUNDLE COSTS


SUM UP TOTAL SHELL (SHELL + CHANNEL + NOZZLE
+SHELL LINING + SHELL ASSEMBLY) COST AND WEIGHT,
TOTAL BUNDLE (TUBES + TUBE SHEETS + BAFFLES +
MISCELLANEOUS + BUNDLE ASSEMBLY)
COST AND WEIGHT

CALCULATE ERECTION LABOR


S FINT OUTPUT




Figure 25









CHEMICAL ENGINEERING EDUCATION


OPTIMIZATION PATTERN


B ABILITY
./ MODEL




C FACILITIES ,- /' CONTROLLABILITY
CONJECTURE CONSIDERATION



RELIABILITY
CONSIDERATION





Figure 26
Computer Design of Equipment

Heat exchanger costs are often calculated from a family of curves giving
cost as a function of heat exchanger area for various materials. Costs were
adjusted for different pressures, tube lengths, bundle types, number of
passes, etc., by means of empirical correlation of actual prices paid. (Fig-
ure 25)

The two examples merely suggest the availability of many other programs
for facilitating design computation. Such programs are prepared by any eng-
ineering organization close to a computer. One can obtain many of the pro-
grams through the AIChE computer program exchange effort or through program-
sharing organizations.
SUMMARY


Industrial design optimization can be defined as the orderly consider-
ation of the pertinent factors which control the possibility of a venture
in the selection of specific equipment and operating conditions to produce
the maximum profitability. Note the word pertinent not all, or all inclu-
sive. In application, techniques must be simplified to the point where they
will make a major contribution in helping guide research in selection of the
proper process arrangement, while, at the same time, not unduly complicate
the procedure at least not beyond the point where the added sophistication
is more than paid for in gain.

If carried to the extreme, quantifying techniques would result in exact
specifications for each component in the system, to insure some optimum tech-
nical performance of the over-all system. Each component would then be de-
signed and built specifically to these specifications. However, the engineer-
ing part of optimization implies consideration of economics as well as tech-
nology so that complete custom designing prices itself out of the market. The
final level is an over-all optimization procedure which carefully analyzes all
factors and locates all maximum and minimum values of cost and profit in the
range being considered.

Finally, this leads to a plea for the systems engineering approach. In
the chemical process which is operated to produce a product for customers, the
static conditions rarely actually are achieved in plant operation. The plant
conditions are continually being adjusted to meet the quality and capacity of
the demands. Start-up and shut-down operations may establish many of the de-
sign criteria, rather than the proposed maximum operating capacity. Therefore,
since the plant operating conditions are continually changing, the plant de-
sign must be based on dynamic considerations in order to maximize profits. It
is becoming too expensive to build a new plant and to experiment with it.
Facts must be known before process is built not after.


Dec. 1962









CHEMICAL ENGINEERING EDUCATION


This means that the optimization pattern (Fig. 26), leading to an op-
timum or "best" plant, must as early as possible and as research proceeds, tie
in with the laboratory program, conjectures and mathematical models of the
process, the proposed facilities, their controllability and reliability.
Engineers have begun to realize that they can no longer think of a pro-
cess plant as a collection of individually designed operations and processes.
It is increasingly apparent that each component of a plant influences all
others in obvious and subtle ways. Some of the subtle influences are readily
found, others can be pinned down only after long, steady experimentation.
This is a challenge to professors of chemical engineering, whose contributions
can, for example, be as follows:

1. Research to define explicitly the dynamics of chemical
engineering operations.
2. Systems and procedures to facilitate the design function.
3. Publications to achieve understanding by industry and
especially of elementary, practical, non-complex appli-
cations of the new techniques.
4. Indoctrination into the college students the fact that
graduates who have been a number of years require "time
out for mental digestion" of the new, more sophisticated
approaches.

In this short review of several industrial design optimization techniques
emphasis has been placed on getting an understanding on what is optimum or
"best," and that cost and degree of optimization must be compared with the
benefits. Further, three types of optimization must be considered for a chem-
ical plant: the process design (or specification of which kind of equipment
is to do what), the project or detailed design (the mechanical features of the
equipment), and the engineering function (how to get the best job done in the
best way). Even though there is increasing use of mathematics and computers,
the individual judgments, which can not yet be relegated to rigorous analysis,
tell us categorically that design of a chemical plant remains to a large part
the Design Project Engineerls or "artist's" personal effort. The challenge
is for the university research to maximize the mechanization of those steps
which lend themselves to mechanization so that the project engineer can op-
timize his time in creating the "best" plant.

LITERATURE CITED

1. Rase, H.F., and Barrow, M.H., Project Engineering of Process Plants,
J. Wiley & Sons (1957). Figure reproduce' whf perm.ssTon or ..y-
right owner.
2. Quigley, H.A., and Weaver, J.B., Ind. Eng. Chem., 53, No. 9, 55A
(Sept. 1961).

3. Smith, W.C., and Giesse, R.C., Ind. Eng. Chem., 53, No. 7, p. 540
(July 1961). Figure copyrighted by permission o-the copyright
owner.

4. S. Ehlers, Ind. En. Chem.53, No. 7, P. 552 (July 1961). Figure
copyrighted y perm ais fn o he copyright owner.

5. S. Ehlers, Ind. Eng. Chem., 53, No. 7, p. 554 July 1961.
6. Van Note, R.H., and Weems, F.T., Ind. Eng. Chem., 53, No. 7, 546
(July 1961). Figure copyrighted by the Amer. Chem. Soc., and re-
produced by permission of the copyright owner.

7. Bartkus, E.P., E.I, du Pont de Nemours & Co., unpublished paper,
"The Systems Engineering Approach to Profit in the Chemical Industry".
(2/8/60), presented to South Jersey, Pittsburgh, South Texas, Col-
umbus A.I.Ch.E. chapters.


Dec. 1962














PROGRAMMED LEARNING IN CHEMICAL ENGINEERING EDUCATION

L. Bryce Andersen
University of Nebraska
Lincoln, Nebraska


The application of programmed learning to the teaching of high school
and beginning college courses has grown at an astounding rate in the last
few years. In September 1962, 122 programs ranging from music to mathematics
could be obtained on a routine basis from the nation's publishers (A 3). In
addition, several hundred more programs are being prepared. Research in pro-
grammed learning is actively pursued in leading colleges throughout the coun-
try (Al, A12).

In view of this intense development of programmed learning in high schools
and colleges, the Education Projects Committee initiated a brief study of pro-
grammed learning as it might apply to teaching chemical engineering. This re-
port is a preliminary survey of the field, with a few suggestions for appli-
cations. Because no programs are currently available in college-level chem-
ical engineering, the report is necessarily rather general and speculative.

As the name implies, programmed learning attempts to more thoroughly
structure written instructional material. The mechanical teaching machine
is perhaps the best known technique of programmed learning; however, the
programmed textbook may offer more promise for advanced college courses. A
third type of programmed instruction utilizing closed-circuit television and
a digital computer is in an early stage of development. One authority (A13)
believes that the simple programmed textbook (because of its low cost) and
the very complex computer-controlled teaching machine (because of its tre-
mendous versatility and capacity) are the two techniques which offer the
greatest promise.

Basic Procedures in Programmed Learning

All programmed learning techniques are designed to present the material
to be learned in a sequence of short steps. Usually each step consists of
one or two sentences and perhaps a figure or diagram. The learner is most
often required to fill in a blank with a word or phrase or to answer a ques-
tion. Most programs require that the answer be written out either on the
program or on a separate sheet. In some cases the answer is fed to the
machine by a keyboard or typewriter. Each step requires the learner to make
only a small increment in learning and clues to the correct answer are given.
As a result, errors are seldom made. After answering an item, the learner
proceeds to the next step, which wasn't visible when he answered the previous
item. Here the answer to the earlier item is given, and another item with a
question is presented. The learner proceeds step-by-step, answering questions
or filling in blanks at each step, until he reaches the end of the program.
Programs range in length from several hundred to a few thousand items. For
example, a programmed textbook on introductory statistics contains 1700 items
and requires 15 to 25 hours to complete. The basic sequential pattern of
programmed learning described here has many variations, several of which will
be described here in a later section.

The proponents of programmed learning claim that it has three major
advantages over conventional teaching methods (A-6):
1. Programmed learning requires continuous, active student participa-
tion. His response to each question gives him practice at each
item, so that each step in the learning sequence is properly learned.
It is difficult to be passive or indifferent when reading and re-
sponding to a program. The continuous demand for answers maintains
student interest.

2. The student learns whether his answer is right or wrong with mini-
mum delay. This tends to make him remember correct responses and
quickly forget erroneous answers. In a conventional classroom sit-
uation a student often waits several days to learn whether his
homework or examination answers are correct. Of course, he may ob-
tain immediate reinforcement of his correct answers from the teacher
in classroom discussion, but not all students can participate. Pro-
grammed learning gives immediate reinforcement to every student --
just as though each student had an individual human tutor.
Educational psychology has shown that immediate reinforcement
facilitates learning. A student will remember an answer which he
knows is correct better than one he is uncertain of. Conversely,
his learning immediately that an answer is wrong encourages him to
forget the wrong response before he has a chance to learn it. Sup-
plying the correction for the wrong response also aids in learning.









43 CHEMICAL ENGINEERING EDUCATION Dec. 1962



3. Each student can proceed at his own individual rate. Fast learners
are not held back by the slow student, as sometimes occurs in the
conventional classroom. Conversely, the slower student is allowed
all of the time he needs.

Do these claimed advantages actually lead to better learning using pro-
grammed instruction? Much research is in progress to compare programmed and
conventional instruction. The results seem to show that programmed learning
is at least as effective as conventional classroom textbook instruction
in some areas (A-13). That is, the student learns as much, and in some cases
more. However, these conclusions cannot be applied to all levels of college
courses, because no studies have been made on advanced college courses.

Since programmed learning requires a sequential presentation of infor-
mation, the material to be taught must be of a type that can be broken Into
a sequence of steps. The most enthusiastic proponents of programmed learning
claim that anything that can be taught can be programmed. More realistic ob-
servers emphasize the need for defining the objectives of the proposed pro-
gram before an attempt is made to write it (A-9). It is necessary to state
very carefully exactly what the program is to teach. This is done by stipu-
lating what the student should be able to do at the end of the program. For
example, statement 1 is a much more approprTate goal for part of a higher
algebra program than statement 2.

1. The student must be able to solve 5 pairs of simultaneous alge-
braic equations in 15 minutes.
2. The student should have developed an understanding of simultaneous
algebraic equations and their solution.

How can "understanding" be measured? Only by requiring an overt response
such as that suggested in statement 1.

The precise delineation of what is to be learned is an indispensable
first step in writing a program. This delineation must state the overt be-
havior expected of the student at the end of the program. If this cannot be
done, there is no point in writing a program because there will be no way to
determine whether the student has learned anything. In high school courses,
which usually emphasize learning of specific facts and techniques, the goals
often can be precisely defined in terms of overt behavior. On the other
hand, such definition is much more difficult in advanced college courses.
It may even be undesirable or impossible in courses where the student is en-
couraged to think for himself and set his own goals. As a result, it may
not be possible to program many advanced college courses. Published programs
include only subjects which are well-structured, such as elementary mathema-
tics, or subjects which require the learning of many facts and rules, such
as English grammar, usage, and spelling. Many advanced college courses do
not seem to fit into these categories.

Most of the effort in programmed learning to date has been concentrated
on high school courses (7th through 12th grade). Much more work is needed
before the value of programmed learning in college courses is established.
However, the dramatic success of some high school programs appears to make
college studies desirable.

Types of Programmed Instruction

Various devices are being developed to present the sequence of items
required in programmed learning. Three major types are teaching machines,
programmed textbooks, and computer-controlled devices. Before these are dis-
cussed, a few comments on the program itself are necessary.

Programs and Programming:

The program is the heart of programmed instruction. Writing a program
is a difficult chore. Although several hundred programs have been written,
there seems to be little consensus on how it should be done. A programmed
primer on the subject gives a few suggestions (A-10). It is also an inter-
esting example of a programmed textbook.

Most psychologists feel that learning is more efficiently accomplished
by reinforcing correct answers than by correcting erroneous answers. As a
result, most programs are written to elicit a correct response. This is ac-
complished by providing sufficient "cues" to the student. The answers to
the first several items on a specific subject may be made obvious with var-
ious cues. The cues are then slowly eliminated in succeeding items until
the student is able to answer questions about the subject without the bene-
fit of cues.

Most programs require written responses. The reader could simply for-
mulate the response in his mind, but many programmers believe that the overt









CHEMICAL ENGIEERINGO EDUCATION


action of writing the response gives a more active role to the reader; and
hence he learns better. Another important use of written responses is in
correcting the program. If an item in the program is too difficult, many
students will answer it wrong. The programmer can then check these responses
and modify the item to assure correct responses.

Not all programmers believe that incorrect responses are necessarily bad.
They may be advantageous if they can be used as a sign that the student needs
more training on the subject of the question. The additional training can be
added to the program by "branching". Depending upon his answer, the reader
is told to go to one of several following items. If his answer is correct,
he goes on to a more advanced item. If he answers incorrectly, he goes on
to a sequence which will correct his mistaken ideas. There may be more than
one branch from a given item.

Branches may also be used to skip items when a quick learner demonstrates
a superior grasp of the material. The other branch then includes more prac-
tice for the slower learner. Branching is used in many other situations
where the programmer wishes to offer more than one alternative sequence of
items. A branch may sometimes be taken at the option of the reader, if he
feels he needs the additional knowledge in the branch. Branches may be of
equal difficulty, where the choice between them is based on an opinion of
the reader, which is neither right or wrong.

Figures, graphs, and long quotations may be included in a program.
They are often placed on separate pages and the student is asked to refer
to them when he comes to a given item. There is no limit to the length of
such additional material. It would be possible to include an entire arti-
cle or book, and then ask detailed questions about the selection using the
programmed items. In this way the teacher can test the student's under-
standing of what he has read.

Devices for Presenting Programs:

The program itself is the core of programmed instruction. The mechan-
ical devices for presenting the program and necessary supplementary material
are really secondary. However, the various devices are sometimes suitable
for different types of programming, so they will be discussed briefly. A
good survey of these devices is given in Reference (A-12).

Teaching Machines are mechanical devices for presenting one item at a
time to the student. They usually require that the student make his re-
sponse (by writing the answer, pressing buttons for multiple-choice questions,
etc.) before the correct answer and the next item appear. Often the machine
is designed to make it impossible for the student to change his answer once
he has seen the correct answer. Teaching machines range from a simple metal
box costing a few dollars to complex devices including motion picture or
slide projectors costing several hundred dollars.

Programmed Textbooks are essentially "paper teaching machines". There
are several methods of programming texts. The resulting books look radically
different from a conventional textbook.

In the horizontally-programmed text, each page is divided into a number
of "frames", arranged from the top to the bottom of the page. Each frame
contains a single item and the answer to the previous item. The reader starts
at the top of the page, reads the first item, and answers it (usually on a
separate sheet of paper if the book is to be re-used). He then turns the
page and reads the frame at the top of the next page. Here he finds the an-
swer to the first item; and the second itemTWs presented. He continues
through the book, reading only the top item on each page. After the last
page, he returns to the first page and reads the second fran e from the top.
He continues this procedure, moving down one frame each time he reaches the
last page. There may be as many as seven frames per page, so the reader will
thumb through the book seven times.

Since the answer to any given question is always on the following page,
the answer is not available to the reader until he has answered the question.
Of course, "cheating" is possible. The reader may look ahead to the answer,
but he gains little because he isn't taking a test. Since there is little
advantage to looking ahead, few readers would do so. It would be somewhat
like looking up the solution to a newspaper crossword puzzle while working
it out. It just spoils the fun.

Obviously, the back of each page cannot be used for the following item,
because it would face the next item and the answer would be visible. As a
result, items are arranged --tuse only the front of each page from first to
last page; and then the program continues from page to page on the back of
each page.

The vertically-programmed text divides each page into a series of frames


De. 1962









CHEMICAL ENGINEERING EDUCATION


Dec. 1962


running from top to bottom. The reader starts at the top and reads to the
bottom. Obviously, the answers to the items are visible on the same page,
so some sort of shielding device is needed. This can be simply a sheet of
paper that is lowered on the page as each item is answered; or it may be a
special plastic cover that serves the same function. The vertical arrange-
ment permits a once-through reading of the book; but it requires some sort
of shield.

A particularly intriguing programmed book is the scrambled text. Here
the reader starts on the first page, where there is a discussion followed by
a multiple-choice question with answers. Depending on the answer he chooses,
the reader is told to turn to a specific page (never page 2). If he chooses
the correct answer, the page he turns to tells him he is correct, and pre-
sents more material and another question. If he chooses a wrong answer, the
page he turns to tells why his answer is wrong and gives remedial work, per-
haps extending through several more questions and branches until it finally
returns to the main (correct) branch of the program. There may be several
wrong answers in the multiple-choice question, each with its own remedial
branch. Obviously such programs may become quite involved, with multiple
branching. Reference (B-5) is an example of a scrambled text.

Generally, programmed texts are less expensive than programmed teaching
machines, because the former doesn't involve expensive mechanical equipment.
For example, a programmed text on elementary electronics costs $26.25. A
machine program for the identical material costs $70, and the necessary
teaching machine costs $700. (Ref. A-3). Texts are more easily adaptable to
branched programs.

Computer-controlled teaching machines offer great promise for the fu-
ture (A-2). Such devices could "tailor-make" programs, taking into account
individual student differences in learning rate, educational background, and
aptitude. The computer could be responsive to each student's needs while it
is handling a large number of students. At each step in the program, the com-
puter may modify the remaining program by considering such factors as

1. Promptness and correctness of the student's answer.
2. Specific errors in the answer.
3. Dpta on the student's previous learning habits; such as reading rate.
4. Personal data, including intelligence, special aptitudes, sex, etc.
5. Nature of the material being studied.
6. Level of student motivation.

Recent developments in computer-controlled teaching machines are dis-
cussed in Ref. (A-4).

An interesting example of the computer-controlled teaching machine is
Plato II,(Programmed Logic for Automatic Teaching Operations) developed at
the Univeriity of IllTnois. ReTding material, figures, and questions are
presented on a television screen. The student types his answers using an
electric typewriter. The answer may be in numbers, algebraic expressions,
words, or sentences. When the student completes his answer he presses the
"Judge" button and the computer judges the correctness of the answer, flash-
ing "OK" or "No" beside the answer on the screen. If the answer is wrong,
the student may ask for additional help by pressing the "Help" button. He
then follows the computer which selects easier related material until the
student indicates he understandsby pressing the "Aha" button. He then returns
to the question missed.

Plate II has been used to teach mathematics and French. At present it
is set up to handle two students simultaneously, but plans call for a larger
number. The number of steps in the "Help" sequences is somewhat limited.
Computer teaching may eventually handle a large class, using only one
computer for storage of the program and control. Each student could proceed
at his own rate with his own television screen and typewriter. The cost of
such an installation would be very high. Writing a teaching program for such
a complex operation would be extremely difficult. The goal would be a pro-
gram which would anticipate every possible student error and would include
sufficient corrective material for even the slowest student. The result would
be a machine which acted very much like a human tutor, in that it would be al-
most completely adaptable to the needs of any specific student.

Programming in College Courses

Of the 122 programs commercially available in the fall of 1962, only
ten are appropriate for college classroom use (A-3). These include two ele-
mentary courses in statistics, two in general psychology, and one each in
seat theory, vectors, probability, chemistry, physics, and "Fortran" computer
programming. In addition another half-dozen programs of post-high school
level are listed. These include a series of short electronics programs pub-
lished by Varian Associates covering capacitors, klystrons, relays, and
switches; one on mathematical logic, and one on basic electronics.


S45









Doe. 1962 CKEMIOAL ENGINEERING EDU CATIO 46



There are certainly more than ten college-level programs in existence,
but many have been locally-developed and are not yet commercially available.

An excellent survey of programming activities presently under way in
engineering courses is given in Reference A-1. The survey shows considera-
ble interest in programmed instruction in engineering courses. Included is
a list of nearly 100 individuals preparing programs of interest to engineers;
but most of these are in the early stages of development. Programs in prep-
aration include elementary courses in chemistry, physics, mathematics, sta-
tistics, mechanics, drawing, electrical engineering, computer programming,
and a few others. Several programs are being prepared in advanced mathemat-
ical subjects, such as matrix theory, Boolean algebra, Laplace transforms,
and vector analysis.

Many of these programs cover a limited subject which would be a small
part of the full course. Undoubtedly some of these programs will be suf-
ficiently well-developed and tested so that they can prove to be of wide-
spread use in engineering colleges,

The survey indicates that many engineering colleges would use programs
if they were available. The interest centers on mathematics, engineering
mechanics, and electrical engineering, but very little interest was shown in
chemical engineering. Programming of engineering subjects is being encouraged
by the Committee on Programmed Instruction of the American Society for En-
gineering Education. They have announced plans for a workshop in programming
techniques for engineering faculty to be held in the summer of 1963 (A-l).

To be appropriate for programming, a college course must have a clearly-
defined objective which can be stated in terms of overt behavior of the stu-
dent. Well-structured subjects such as elementary mathematics.and natural
science seem to have well-defined objectives. On the other hand, advanced
courses in engineering design do not have specifically-defined objectives
and could not be programmed.

It is not necessary to program an entire course. Only those parts which
are well-structured need be programmed. The teacher can use programs along
with traditional teaching techniques. Fot example, programmed Fortran in-
struction could be inserted in the beginning of an engineering design course
which required solution of design problems on the computer.

Programming, even with branching, seems to force the student into a
highly-structured pattern. In advanced engineering courses emphasizing in-
dependent thought such structuring often would be undesirable.

Engineering has emphasized the application of physical principles to
the solution of complex problems. Although the student may learn the physi-
cal principles by programs; it appears that the teaching of the solution of
complex engineering problems is often too unstructured to be appropriate
for programming.

What courses in a typical chemical engineering curriculum could be at
least partially programmed? Any attempt to answer this question is, of neces-
sity, pure speculation. Elementary thermodynamics would appear to be suf-
ficiently defined to permit programming. Some of the basic concepts and
definitions of mass and energy balances might be programmed, although the
more complex balances could not and should not be. Many of the basic con-
cepts of stage and rate operations could be programmed, but it might be dif-
ficult to integrate the programs with non-programmed material and homework.
Similarly, basic concepts in kinetics and process dynamics might be programmed
although most of these courses would not be.

In basic courses which are prerequisite to chemical engineering, pro-
grammed instruction appears to have many applications. Most of freshman
mathematics, chemistry, and physics could be programmed. Reference (B-6) is
a college physics program consisting of 12,000 items. An experimental fresh-
man chemistry program is given in Reference (B-10).. Basic engineering mechan-
ics can be programmed. (An experimental program in Kinematics is given in
Reference B-7). Elementary organic and physical chemistry appear to be ap-
propriate for some degree of programming. An excellent program for basic in-
struction on the slide rule is available in experimental form (B-9).

A survey of several of the major publishers of engineering books indi-
cates considerable interest in programmed college texts, but there are very
few definite announcements of programmed engineering books to be published
in 1963 or 1964. No programmed chemical engineering texts have been announced.
The problem would appear to be one of finding authors sufficiently familiar
with engineering and with programming.









CHEMICAL ENGINEERING EDUCATION


Interested engineering teachers might try writing a program for a sec-
tion of one of their courses to evaluate the utility of programmed instruc-
tion. Guides on programming techniques are available (for example Ref. A-8,
A-10), Unfortunately, the techniques of programming are not sufficiently
developed to make it either easy or enjoyable.

Conclusions

Most effort on programmed instruction has been devoted to high school
courses. AS a result, it is impossible to state whether programmed in-
struction can be successfully applied to college engineering courses. No
conclusions can be drawn until extensive experimental studies have been made
in engineering courses. In addition, much more research is needed to clarify
the psychological principles of programmed instruction, so that the practice
can be .put on a sound theoretical basis.

Programming is still an art. Although the various schools of program-
mers can cite psychological principles which they believe support their meth-
ods, few parametric studies to investigate their claims have been made.
Greater understanding of programming methods should precede any major effort
to program engineering courses. Methods developed for high school use are
not necessarily the most appropriate for college courses. Some educational
psychologists have questioned the criterion that requires most student re-
sponses to be correct. Perhaps more error would be desirable. One study
(A-6) indicates that the activity of filling in blanks or writing out the
answer may not be necessary. Possibly the value of a program is in the care-
fully-developed sequential arrangement of the important concepts to be
learned.

Most of the studies to date have compared independent programmed learn-
ing with instruction using conventional teAs and classroom discussion. Pos-
sibly studies using both programmed textbooks and classroom discussion would
show that the combinia-on is superior to either above. There is no need to
substitute programmed texts for other techniques of teaching. In engineering
courses particularly, a combination might be quite effective. A programmed
text could be substituted for a conventional text in, for example, an exist-
ing thermodynamics course. Class discussion and homework could continue as
before.

Can programmed learning be used in chemical engineering education? It
is too early to tell. The final answer to this question can be obtained
only with the active participation of engineering faculties. Engineering
professors must work closely with educational psychologists in developing and
testing suitable programs. Although such cooperation would be a novel exper-
ience for both groups, it could prove very beneficial to teaching in chemical
engineering.

The potential of programmed learning is neatly summarized by R. D.
Patton (A-11); "The enfant terrible of the moment is the teaching machine.
No one can say for sure wtatFl-Inaor adult it may grow up to be. Meanwhile,
it has struck fear into the hearts of all but the bold and perhaps the fool-
ish. The fear is that it may subvert the teaching art into a slick kind of
game-playing, or at best so attenuate the humane aspects of the teacher-stu-
dent relationship as to cause the loss of the contagious joy of discovery
which has so often sustained the intellectual life in the past. Yet there is
much routine learning which must precede and accompany discovery; and it is
quite probably that auto-instruction devices may perform this function as
well or better than the teacher in the flesh. Fortunately, many of the older
and saner heads in this area of instruction are urging caution and avoidance
of extravagant claims until more experience is gained and more research com-
pleted. If good sense can carry the day, there is a very real possibility
that the self-teaching device, whether it is a true machine or a programmed
textbook, may relieve the human teacher of much drudgery and permit even more
intimate contact between the older and younger learner".

REFERENCES

A. Books and Articles Giving Information on Programs, Programming,
and Programmed Learning.

A 1. American Society for Engineering Education, Committee on Programmed
Instruction "A Report on Programmed Instruction" J. Engr. Educ.
53:117 (Oct. 1962)

A 2. Bushnell, D. D. : "Computer-based Teaching Machines", J. Educ. Res.
55:528 (1962)

A 3. Center for Programed Instruction: "Programs, 162- A Guide to Pro-
grammed Instructional Mpterials" U.S. Government Printing Office,


Dec. 1962









HOEM1CAL ZN IRINU 0 OCATIZi


Washington (1962) This book lists all programs commercially available
in September 1962. It includes a description of each program and sma.
ple pages. The index lists several additional programs of interest
to engineers. Cost, $1.50.
A 4. Coulson, John (ed.) "Programmed Learning and Computer-based In-
struction", John Wiley and Sons (1962)
A 5. Deterllne, W. A., "An Introduction to Programed Instruction", Pren-
tice-Hall, Englewood Cliffs, N.J. (1962) A general introduction in-
oluding an illustrative program. Paperbound, 131 pages.
A 6. FPeldhusen. J. F., and A. Birt: "A Study of Nine Methods of Present-
ing Programmed Learning Material", 3. Edue, Res. 5:461 (1962)
A 7. Lumadaine, A. A., and R. Glaser (ed.) "Teaching Machines and Pro-
grammed Learning National Education Association, Washington (1960).
This 724-page book collects the important articles in the field up to
1960. Cost: $7.50
A 6. Ly saght, J., and Williams: "Guide for Programmed Instruction",
John Wiley and Sons, New York. To be published February 1963.
A 9. Mager, R. F.: "Preparing Objectives for Programmed Instruction",
Pearon Publishers, San Francisco (1962)
A10. Markle, S.H., L.D.Eigen, and P.K.Komoski "A Programed Primer on
Programing" Caenter for Programmed Instruction 365 West End Ave.,
New York (1961) An easily-read programmed text giving an intro-
duction to principles of writing programs. Cost: $2.00
All. Patton, R.D.: "Teaching" J. Higher Educ. 33:277 (1962)
A12. Stolurow, L.M.: "Teaching by Machine" U.S.Government Printing Of-
fioe (1961) A comprehensive surr of the field. It includes
descriptions of various machines, a discussion of the principles
and practice of programming, a summary of research findings, and an
extensive bibliography. 173 pages. Cost: $0.65
A 13. Stolurow, L.N,: "Implications of Current Research and Future Trends",
J. Educ. Res. 55:519 (1962)
B.. A Selection of Comnercially-Available Programmed Textbooks of In-
terest to Engineering Teachers
B 1. Basic Systems, Inc.: "Vectors" Appleton-Century-Crofts, New York
High school or beginning college level; 495 frames; $2.75
B 2. Colman, H.L., and C.P. Smallwood: "An Auto-instructional Intro-
duction to Fortran Programming" McGraw-Hill Book Co., New York
College Level; 1000 frames $3.95
B 3. Evans, J.L., and L.H. Home: "Introductory Statistics" TMI -
Grolier, Hew York.Advance high school beginning college level;
1700 frames; $10.00
B 4. General Education staff: "Probability Models of Random Processes"
General Education, Inc. Cambridge, Mass. College level; 800 frameat
$15. 00o
B 5. Hughes, R.J., and P. Pipe: "Introduction to Electronics" Doubleday,
New York.
Intended for self-instruction in electronics for people with
little mathematics background. Not a college text, but an interest-
ing example of the scrambled textbook. 14.95.
B 6. Joseph, A., and D. Leahy: "Programed College Physics" John Wiley and
Sons, NHew York. College level, but at a low mathematical level;
12,000 frames; $15.00
B 7. Leahy, D., "Kinematics" Center for Programed Instruction, New York
(Experimental Edition)
B 8. MoFadden, M. "Sets, Relations, and Functions", McGraw-Bill Book Co.,
New York. High school and college level; 1150 frames; $3.50.
B 9. Morley, P. "Slide Rule Operation" Center for Programed Instruction,
New York (Experimental Edition)
B 10. Young, J.A. "Selected Topics, Preshman College Chemistry" Available
from author, Kings College, Wilkes-Barre, Penna; 5000 frames; $8.00.


Dee. 1962












4 important publications from McGraw-Hill


1. MOMENTUM, HEAT, AND MASS TRANSFER
By 0. 0. BBMWBTT, Manager, Process Research and Development, The Ltena
OCpoatn eand z ,. z'iRS, Purdue University. MoGraw-Rill Series In Che.leal
Qginearins. 69; pages, $13.50
An outstanding new modern introduction to chemical engineering. The authors
have set forth the fundamentals of momentum, heat, and mass transfer and
have applied them to engineering problems, particularly those in the process
industry. The manner of exposition makes this book especially suitable for
an undergraduate text in engineering. Luold writing results in comprehensive
yet concise coverage of the fundamentals of momentum, heat, and smam transfer,
with a number of applications in engineering in general, and ehemioal ongin-
o**rig in particular. A discussion of the unit operation. is inelu edi but
the order of the subleot matter has been determined more by the strutaure of
the theory than by the usual division into unit operations.

2. CHEMICAL ENGINEERING CALCULATIONS
By 33B3s s. BR3StL, Stevens Institute of Teohnology, and MERRMA BIBBER, Easo
Resaroh and Bna uaering Company. Moara*Hlll Seriea in Chenaoel 3laneerin'l
441 pages $9.0
fundamental concepts, rather than technology and empirioiam, are stressed
throughout. Engineering techniques suph as graphical method, approxinations,
trial and error of the calculus is made throughout. A unique feature of tiLs
book is its inolusiv eoverage of thermodynamies, material balances, phase
equilibrium, and oheloeal reaction equilibrium. The first nine chapters
provide a rudimentary treatise In problem solving techniques, as illustrated
by material balaoee ealo4uations. Next a detailed exposition of the First Law
is offered, and finally illustrations involving all previously 4dieoused
disciplines are given. The Second Law is used, and derived in the Appendix,but
is not discussed in detail.

3. DIGITAL COMPUTATION FOR CHEMICAL ENGINEERS
By LZON LAPI S, rinoeton Univeraty. MoGraw-Hll Series in chemical EnAin-
eering. 1432 pages, 11.50
Dioeasee various areas of digital computer mathematics of importance to the
obemeial engineer. The increasing emphasis on complex alculations for ahea-
ieal engineering analysis has led to the increasing use of digital and analog
computerss. This text is designed specifically for the teaching of digital
teeo niqu in the chemical engineering curriculum.

4. ENGINEERING AS A CAREER emod edition
By RAPM SuI Stanmford envelaty. 14S ) pagee, 4.95(0 thC), $3.50(paper).
ThiA introduetery orientation and problem text in engimmering provides a
aomprehesivTe and factual picture of an engineaeres duties, qkalifiostiona and
training needed, and career opportunities available. The author explains the
engineering profession in terms o funations as well as branches. New chapters
indl*edt "Adjuestent to College* and "Vow To Be A Better Student". lew ex-
amples, illustrational material updated.


DEPARTMENT 01-03
McGRAW- HILL BOOK COMPANY
330 West 142nd Street New York 36, N.Y.




Full Text

PAGE 1

' ,: ,. ,. ; .. ; .! I I l I .. ; I THE .. .. ' t .. I .. -~ : I '' ,, .. . I ... ,._-! ,, I . ,:. \ .. .. : .. .. .. 'I 'I ,. I ,' CHEMIC AL ENGINEERING DIVIS ION AMERICAN SOCIETY FOR ENGINEERING Deoe ber 1962 -, . f. :i; ': ., '. \.,. ... ,-,. I ,. ,.. \ I ,I.\ .. '' ', ,, ,; ', ..... .. ,, . '.'r "' 7 EDUCATION .. ,, .. .. ,. '' ... \ ~" I ,, r :

PAGE 2

y I CHEMICAL ENGINEERING SERIES Consulting Editor: PROFESSOR CHARLES R. WILKE, University of California at Berkeley. COMPUTATION OF MULTISTAGE SEPARATION PROCESSES by DoNALD N. HANSON and GRAHAM F. SoMERVILLE, both of the University of California at Berkeley, and JoHN H. DUFFIN, U.S. Naval Postgraduate School, Monterey, California. The first book to present the mathematics of stagewise separation processes in the form of Fortran computer programs used to solve separation problems in vapor-liquid processes and liquid-liquid extraction. It will prove invaluable as a text for advanced courses in separation operations. PLOW OF FLUIDS THROUGH POROUS MATERIALS by R. E. CoLLINs, University of Houston. A unified treatment of all aspects of the flow of fluids through porous materials. This book is valuable to petroleum engineers, chemical engineers, civil engineers, and soil scientists. 1961. 280 pages. $12.50 AN INTRODUCTION TO CHEMICAL ENGINEERING by CHARLES E. LITTLEJOHN and GEORGE F. MEENAGHAN, Clemson College, Clemson, S.C. This book emphasizes the fundamentals upon which chemical engineering theory is based. It contains a wealth of material on the professional aspects of the field unavailable in other standard texts. 1959. 288 pages. $6.50 PLUIDIZATION AND FLUID-PARTICLE SYSTEMS by FREDERICK A. ZENZ and DoNALD F. 0THMER, both of Polytechnic Institute of Brooklyn. This comprehensive work provides a wealth of data on fluid particle operations answering problems common to process ind us tries. 1960. 523 pages. $15.00 DI STILLATION: PRINCIPLES AND DESIGN PROCEDURES by ROBERT J. HENGSTEBECK, American Oil Company. Here is all the information needed to design any distillation column for which vapor-liquid equilibrium data is available or can be estimated. New material is presented on methods for calculating the ''splits'' of the ''non-distributed'' components in multi component distillations, and for minimizing trial calculations for flash vaporizations. 1961. 380 pages. $11.50 FILTRATION by GEORGE D. DICKEY, P.E., Consultant. A modern account of solid-liquid separation in wet processes: water, industrial products and wastes. It offers a comprehensive study of filtering, including a summary of mathematical theorie s and formulas and a short history of filtration development by gravity, vacuum pressure, and centrifugal force. 1961. 364 pages. $12.00 Two Ot her Outstanding Chemical Engineering Books RIEGEL'S INDUSTRIAL CHEMISTRY, New Sixth Edition Edited by JAMES 0. KENT, West Virginia University, with the support of a large number of collaborators. From materials handling to product application, this book offers a handy cross-section presentation of current practices in the major chemical and process industries. A large number of collaborators, all recognized experts in their fields. contribute to make this one of the most authoritative works of its kind. 1962. Approximately 950 pages. HANDBOOK OF VECTOR AND POLYADIC ANALYSIS by THOMAS B. DREW, Columbia University. A probing treatment of the vector and tensor concepts necessary in fluid dynamics. diffusion theory, electro magnetic theory, and heat transmission. It is invaluable as a text for engineering students at the advanced undergraduate and graduate levels. 1961. 112 pages. $5.50 REINHOLD BOOK DIVISION 430 Park Avenue/New York 22, New York

PAGE 3

CHEMICAL EHGlJE~IBO EDUCATION December 1962 Quarterly Journal FUbliahed by the Chemical Engineering DiTiaion A111erican Society for Engineering Education CONTE?r.rS Should We Abandon Chemical Technology by Charles E. Dryden Rew Design Methods in Chelll1.cal Engineering~ The Synthesis or control systems, by Joel o. Hougen Instrumentation in Design, by Kenneth A Otto Design Data and the Role or the Pilot Plant, by w. L. Larcamp Industrial Design Optimiz\tlon, by Edward P. BartkUa 1 19 23 Progrtt,mn~d Learning in Chemical Engineering Education, by L. Bryce Ande rN.D 4-2 Chemical Engineering Division American society tor Engineering Education ott1cer 1962-1963 Max s. Peters (Colorado) Chair aatn Joaeph J. Martin John B. West (Michigan) (Oklahoma State) (Loui1Tille) Vice Chairman Secreta17-!reasurer General Council M. H. Chetrick CHEMICAL EIGIIEERING EDUCATIOK, Journal ot the Chemical Engineering Division, American Soo1et7 fol!' Engineering Education. Albert H. cooper, Editor Publiahed Quarterly, in Maroh, June.I. September, Deoeabe Publication Ottice: uniTersity of ~onneoticut P.O. Box 445, Storr1 1 Connecticut Subaoription price, tz.oo per year.

PAGE 4

SHOULD WE ABANDON CHEMICAL TECHNOLOGY? Charles E. Dryden, Professor of Chemical Engineering The Ohio State University Several years ago, an article appeared in CHEMICAL ENGINEERING EDUCATION by Professor L.B. Anderson (1). Its title was ''Is Unit Operations a Dirty Word?'' His conclusions were that modernized concepts of unit operations would indeed be around for~ long time. In the light of today's emphasis by many educators on engineering science,it would seem even more appropriate to ask the same question about technology, in particular, chemical technology. Technology is defined herein as applied science. The prefix ''chemical" simply denotes the application of scientific principles by engineers to solving the problems of the chemical industries. The net result is chemical engineering synthesis--the putting together of the many facets of science and engineering principles to guarantee performance in keeping the chemical industries in the forefront to ~eet human needs. This is chemical technology in its broadest sense. It encompasses research, development, design and systems analysis, manu facture, marketing and sales. All of the basic fundamentals learned in a modern chemical engineering curriculum are utilized to a degree, including chemistry, physi cs, mathematics, thermodynamics, kinetics, unit operations, and economics. In its narrowest sense, chemical technology can be archaically described as a highly descriptive tale of the birth, life, and often death, of many a chemical industry. It is obvious that there is a vast difference between these two ex tremes as to what constitutes chemical technology and therefore how it should be taught, if at all, to chemical engineerin& students. If we accept chemical technology in the broadest sense, there should be no doubt in the educator's mind that this type of material must be taught to chem ical engineers. The mechanics of teaching chemic~l technology have been debated for years. An earlier commentary on this subject has been given by Koffolt (2) in 1938 and by Withrow (5) as far back as 1911. Since chemical technology rep resents a synthesis concept involving a number of basic principles, we at Ohio State have long felt that it is logical to place an intecrated chemical technol ogy sequence in the senior year. This is supported by the recommendations of the Grinter report, p 14, issued in 1955 by an ASEE Committee on Evaluation of Eng1neerin& Education (2). One of the chief questions is whether the student should vicariously exper ience this vast realm of chemical technology by studying in depth the patterns of numerous chemical industries or whether he should personally wrestle with several realistic and hopefully new problems of the chemical industries and arr1Te at his own solution. While there is less room for ar1ument amon& chemical en11neerin1 teachers on this latter method, the study of industries is certainly controversial. It 1s our opinion that both methods should be used and intecrat ed in such a manner that the case history--vicarious experience material is in jected at the start of a technology sequence in a painless and interesting manner. This combination and timing works to good advantage in an over-all years proiram of study in chemical technoloiY Table 1 summarizes the course sequence in chemical technology at The Ohio State UniYersity. Durini the Fall QuRrter, a comprehensive survey of the chem ic al process industries (Ch.E. 761) is introduced. This controvereial teaching concept will be discussed in detail later. coupled with the background type of course is Chemical Engineering Economy (Ch.E. 760). The principles of economic balances, time Talue of money, and profitability analyses are typical of the subject matter taught. The laboratory work in Chemical Engineering Economy con sists of several comprehensiTe economic an a lysis problems,whereas in the process study course about 25% of the laboratory time is spent in plant Tis1ts and the balance in library research and reporting. In the Winter Quarter, the process development course, (Ch.E. 770), is taught on an informal basis with the students giYen a typical chemical process study. The sequence includes library research, laboratory and pilot plant experimentat ion, prel1118.ry process design, and economic a.nalysis. Several methods are used, depending on the type of problems and size of class. The students work in groups of 3-5 on one of seTeral related processes or es an entire group on one problem. In the latter case, an industrial research and development group is simulated with assig11111ents rotated periodically throughout the quarter. This particular method deTelops management and communications skills as well as tech nical specialization since it is impossible for each student in a large groupto follow completely the work of others in an over-all coordinated project. IndiYidual solution of the lectures on use of computers Quarter design sequence. AIChE Student Contest Problem, in optimization studies round 1 Ch.E. 790, plus out the Winter ...

PAGE 5

Dec. 1962 CHEMICAL ENGINEERING EDUCATION 2 TABLE 1 CHEMICAL TECHNOLOGY SEQUENCE SEHI OR YEAR CHE?-UCAL ENGINEERING DEPT. OHIO STATE UNIVERSITY ~uarter, Course Number and Title A ALL SUARTER Ch.E. 760 Chem. Engr. Economy Ch.E. 761 Chem. Engr. Processes B. ~NTER QUp.RT~R c. Ch.E. 770 Chem. Engr. prooess DeTelopm.ent Ch.E. 790 AIChE Student Contest Problem and Systems Analysis P~INQ 9tUARTE~ Ch.E. 772 Chem. Engr. Process Design Engr. Draw. 755 Plant Design Ch. E. 791 Special Project Problems Investigations credit Hours 3 3 4 2 3 3 Lecture Hrsfwk. w a 2 2 2 after 30 de.,period l l Laboratory Hrs/Wk 2 (computation) 2 (25% on plant trips) 12 (.50% experi mental) 100 hrs over 30 day period 6 6 15 (0-90% experimental) A chemical engineering design seq uence is gi Yen in the Spring Quarter. The process design course {Ch.E. 772) starts with a new problem for the purpose of teaching optimization methods of process design. Digital and analog computers are used to aid in the solution of a relatively complex problem where many basic scientific, engineering, and economic principles influence the results. The plant design course {Engr. Draw. 755) covers the principles of plant layout and auxiliaries design, again using another new problem. The special projects problem (ch.E. 791) is usually conducted as an indi vidual assignment to the student by one of the professors. The scope varies widely and may run from a design project with little experimental work to the opposite extreme. The criterion in each case ls to have the student solve some challenging problem. In summary, the chemical technology sequence presents the students with five major situations which must be resolved on a professional basis. In addit ion, they have acquired a well-rounded knowledge of the chemical industries and a good foundation in the key area of economics. This sequence of courses would represent about 12% of the usual 4-year curriculum in chemical engineering. This percentage is within the proper scope as recommended by the Grinter report (2), page 22. I should like to proceed next to the truly controTersial technological background course as encompassed in Chemical Engineering Processes (Ch.E. 761). This type of oourse has (or had) many titles such as industrial chemistry, chem ical technology, and chemical process industries--to name a few. The philosophy of our present course can be illustrated best by first giTing some past history. Before a major modernization of our chemical engineering curriculum 1n l9$9, this course had been taught for many years as a two-course sequence of three credit hours each in the last two quarters of the senior year. Two text-reference type books with a total of 2074 pages were used for study. Eight inspection trips were taken to nearby plants with reports to be prepared. Needless to say, the course was difficult to teach, eTen for professors with a great deal of prior industrial experience. The rearrangement of the technology sequence to the present scheme shown in T r ble 1 reduced the time allowed for acquisition of technological background to one 3-hour course at the start of the sequence. There was some debate as to whether this reduction in hours could be best ab sorbed by studying only a few chemical processes in depth or by taking a broad brush approach. The latter was chosen principally because all of the other chemica1 eng1neerlng courses were quantitative in content and the broad qualita tiYe viewpoint often needed by engineers for prospective was lacking. A second reason for a wide look s.'t the chemical industries was to incorp orate and orient the students background and preparation in chemistry. No other course in the chemical engineering curriculum can better accomplish this aim. In an analogous manner, the teaching of transport processes serves to utilize the s tudentst formal training in mathematics and physics. The importance of

PAGE 6

3 CBEMI CAL EIGllE&!RIBG EliEJCATIOH TABLE 2 CHEMICAL TECHNOLOGY OUTLINE SERIES TABLE OF CONTENTS I ORIENTATION A INTRODUCTION B. CHEID.CAL AND ENGINEERING LITERATURE c. PLANT INSPECTION TRIPS D. CHEMICAL INDUSTRIES FACTS AND FIGURES E. GENERAL PRINCIPLES APPLIED IN STUDYING AN INDUSTRY l. Chemistry 2. Thermodynamics 3. Kinetics Dec. 1962 4. Chemical Engineering Unit Operations and Unit Processes S. Process and Mechanical Design 6. Economics II. INORGANIC CHEMICAL INDUSTRIES A, SULFOR AND SULFURIC ACID B. FUEL AM'D INDUSTRIAL GASES C. NITROGEN D. WATER E. ELECTROCHEMICAL F. CHLORALKALI G. PHOSPHOROUS H. NUCLEAR RAW MATERIALS I. HIGH ENERGY CHEMICALS AND FOELS III. NATURAL PRODUCT INDUSTRIES A. UNIT PROCESSES AS A STUDY BASIS B. OILS, FATS, AND WAXES C. SUGAR AND ST ARCH D. PULP AND PAPER E. PETROLEUM IV. SYNTHETIC ORGANIC CHEMICAL INDUSTRIES A. PETROCHEMICALS A FRAMEWORK FOR STUDY OF ORGANIC PROCESSES B. CHEMICALS FROM 01 ALIPHATICS C. CHEMICALS FROM C2 A LIPHATICS D. CHEMICALS FROM C3 ALIPHATICS E. CHEMICALS FROM c4, C5 ALIPHATICS F. CHEMICALS FROM AROMATICS G. MISCELLANEOUS V. POLYMERIZATION INDUSTRIES A. F'Olf DAMEN'l' ALS B. TECHNOLOGY l. Thermoplastic 2. Thermosetting J. Elastomera 4. Fibers chemistry is pointed out by the AIChE Committee on Dynamic Objectives for Chem ical Engineers 1n their report ( 5 ) from which I quote: "In the pa.st, the educa tion or chemical engineering undergraduates has been unique in that there has been extensive training in the parent science of chemistry. MPny forces are now tending to eliminate thfs feature of chemical engineering, but this committee bel1eves implicitly that instead, it ought to be carefully preserved and enhanc ed." What better place for the enhancement of chemistry than in the teaching of chemical technology? A third reason for the over-all viewpoint was to help the students solve their tiye technology problems during the senior year with some baekground asto what others had done in the past. This is much the same idea as a research chemist doing a literature survey before making final plans for his own reaearoh program. AS anticipated, we were unable to find ~uitable texts for such a short oourse which meets for a 10 minute quiz plus a 40 minute lecture twice a week for eleYen weeks. To streamline the study program, only the important features ot each chemical process industry were to be covered, not the complete details. A set of notes was prepared 1n essentially outline form to achieve this aim. The table of contents 1s listed in TAble 2. Each industry was discussed in the top ical manner listed next:

PAGE 7

Dec. 1962 l. 2. 3. 4. CHEMICAL EKGlSEERING EDUCATIOW Physical properties of raw materials and products Consumption pattern Methods of production Chemical reaoti ons Process deacrlptlon with flow sheet MPjor engineering problems including thermodynamics, kinetics, process and equipment design, corrosion. Economics 4 In general, in detatl. only one or two important One industry was assigned processes from an industry were for study each lecture period. developed These notes were nearly completed with man~ sections being sent to appropr.1ate industrial companies who have graciously offered comments tor revision. With this ooneiae and direct study approach to each industry on the list, the student's time is not diluted in trying to get details via the more expansiYe reference book approach. Even with this new format, the student cannot possibly absorb all or the material with its implications. RoweTer, the results of obtam ing a broad viewpoint of the ~pproach to problems in the chemical industries haTe been gratifying in performance aehieTed throughout the remainder of the technology aeguenoe. An added advantage accrued in giYing the student a better sense of balance in interviews and a final choice of jobs. Teaching in this style was easier and more interesting. With the knowledge that the students had studied the fundamental points in the outline guide of a particular industry assigned for that day, the instructor felt free to discuss in class some new innovation or recent engineering break-through related to that industry. Aa a result of streamlining the study of chemical technology background material, additional time ls aYailable for 1nd1Yidual project problems. One assignment is doing library research and reporting on two new processes. One or these processes 1a often carried on into process deTelopment the following quar ter (Ch.E. 770). Another part of the laboratory requirements of the course is to make three chemical plant inspection trips as a group and then write up indi Yidual reports. The course is further used as training in the methods of communications. Report writing and public speaking were remoyed from the new curriculum in order to substitute more mathematics and hUlllan1t1es. consequently, written reports in the Ch.E. 7 6 1 course are severely graded and oral presentations are given with the aid of a tape recorder for speech training. SUMMARY The definitions and teaching or chemical technology haYe been presented in a broad, yet penetrating and modern style. The learning or necessary background material to orient the students in a chemical technology sequence has been streamlined to make an up-to-date, fast-moving, and interesting approach to this study of the processes and problems of the chemical industries. When we consider that over 80% of our B.S. degree men go directly to the chemical industries where they immediately encounter many phases of chemical technology. why not prepare them in adYance with a mature and seasoned approach to their new problems? This is the aim of a full year of integrated coursework in chemical technology. This is the z,eason we should neYer abandon nor eTen de crease our teaching efforts in chemical technology for the lure of engineering science per se. If we do, we will soon giTe up the heritage our predecessa~ haYe so stoutly made tor us in best aerylng m a nklnp through the combining of chemist.ry and engineering. REFERENCES l. Andersen, L.B., CHEMICAL ENG. EDUCATION, Trans. of Ch.E. D1T., ASEE, P 13 (1960). 2. Koffol t, J .H. "The Teaching or Industrial Chemistry,'' Paper Presented at 46th Annual Meeting of SPEE, College Station, Texas, 1938. 3. Withrow, J.R., "Points of View in Teaching Industrial Chemistry," J.A.c.s., 33 624 (l9ll). 4. "Report of the Oommlttee on Evaluation of Engineering Education," L.E. G~inter, Chairman, ASEE, Urbana, Illinois, (June 1$, 1955). 5. "Dynamic Objectives for Chemical Engineering," CEP, 57 (10), 69 (1961).

PAGE 8

NEW DESIGN METHODS IN CHEMICAL ENGINEERINGTHE SYNTHESIS OF CONTROL SYSTEMS Joel o. Hougen, Technologist Engineering Department Research and Engineering Division Monsanto Chemical Company st. Louis, Missouri In recent years chemical engineers have become aware of the success which others have achieved in the design of high performance systems. I am referring especially to those electromechanical designs broadly referred to as servomeeh e.hi.ama where special attention has been focused on dynamic behavior and the suppression of transients attising from disturbances. Chemical engineering faculties have responded to the recognized need and are including in their curricula various amounts of control-oriented subject matter and are fostering closer affiliation with their counterparts in other departments in order to reinforce their own efforts. As a consequence more and more chem ical engineers, especially those with graduate training, reach industry with ex cellent backgrounds in education and training in the new control-oriented dis ciplines. I wonder, however, if the teachers get much feedback from their efforts. Perhape you may question if th is special training is being used, and if so to what ends. You may like to know the nature of the problems encountered, the techniques used to attack the problems, and the degree of success in achieving a solution. Moat certainly the demand for chemical engineers with a control engineering speciality is increasing. Recruitment of personnel of this caliber becomes in creasingly difficult, and there is healthy competition between industry and col leges. The ability of industry to convert the most recent theoretical develop ments into reality at an acceptable rate often is responsible for talented per sonnel returning to academic ranks. However., some notable accomplishments ha.ve occurred. The most spectacular is the increased use of computers for on-line, openor closed-loop control. In all, a total of 159 control computers have been sold to date, about one-third to chemical, petroleum, and allied industries. (1) For such work chemical engineers with strong theoretical inclinations combined with aptitude for moQel building and programming are 1n demand. The objective is to design a scheme of plant man agement which will improve and ma1n tain a desired optimum. To date I do not know of a case where the basic concepts of production have been altered because of the availability of the computer. The process design has ; usually been quite conven tional. Special measuring and control components may be added to accommodate the computer, but usually the computer is superimposed on the process to g ive over-all guidance to a more-or-less conventionally designed control system. The major deterrent to effecting new designs lies in the lack of understand ing of the complex prooessea found in chemical engineering. Theoretical formu lations have on one hand been sometimes over-simplified or on the other made ~veP-compl1cated -in each instance leading to design procedures having limita tions in utility. One way of eliminating the above dilemma is to use experimen al data, but this suffers because of the time and expense involved in acquirin g the desired information, and indeed may be even more limited in utility unless suitable pilot plants or prototype facilities are already available. A happy combin a tion of both appears to be the sound approach. For the last five years I have been heavily engrossed in experimental work, both with small and large scale processes. The plant work has usually been directed toward the solution of a problem on an existing process. Thus, the pro cess design has been already established in the large, only minor changes being feasible. However, auxiliary systems have required redesign, these being the systems necessary for control AS an illustration of a typical control system synthesis for an existing plant, I shall describe an actual case. All steps will be described, including the experimental work and terminating with the recommended scheme for control. An analysis of the problem will be made followed by a description of the methods of testing and data collection, reduction, and interpretation. The reasoning used to arrive at an acceptable control configuration will be presented which will be supported by theoretical deductions. Finally, it will be shown where the experimentally determined information is utilized in order to arrive at the best design. Proprietar~ considerations prohibit revelation of the details of the pro cess, but this need not detract from the value of the example. I

PAGE 9

Dec. 1962 CHEMICAL EITGIIEERIKG EDUCATION 6 ~ 1he function of the process was to remove a con d ensable product from a gaseous mixture g enerated in a react o r and t o di sc h arge n o ncon d ens a ble gases into a common manifold for processin g elsewhere. In this pa rticul a r process it was important to maintain the pressure in t h e reactor at a very precise value; name ly, within about t 1/4" of water around some established level. It was the ob jective of the study to design a control system for this purpose. The problem was made especially difficult because of the uncontrolled dis tur b ances associated with the reactor. Gas evolution, and hence pressure, with in the reactor was highly dependent upon the energy absorbed therein and ener g y supply was subject to random variations. In addition, transitory chan g es in the e ffluent gas temperature could occur, these giving rise to pressure chan g es. The gas, approximately 15, 00 0 cu. ft./min., was conducted to a large, dis tributed-type condensing unit, thence to a scrubber, and was finally removed by compressors from which it was discharged into a main serving as a manifold for other similar units. Since the main pressure was subject to random variations, disturbances from this source could also enter the system. These, however, were always greatly at tenuated because of the virtually constant displacement characteristics of the compressors and the relatively high pressure gain across them. It seems invariably true that the general configuration of a control scheme is conceived by logical reasoning, using qualitative performance data on criti cal processing items. However, once a scheme is selected the detailed descrip tions are used to compare performance of one scheme with another or with some established criteria. Thus, mathematical formulations do not necessarily con tribute to the conceptions, but are necessary for veriflcation. This case was typ~c al in this respect. In this instance some basic requirements for satisfactory operation are ob vious: 1. For each specific mass flow rate a given pressure gradient between reactor and compressor suction must exist in order to assure the appropriate flow~ r-... ....C ,~_._. _______ __ .., ~ ~ r---J r I I I I I _.,,_ i I I : G n I io-:..J f I I I I I I I I I r--, r. : G i ~ I -------t __ I I I I I I I Gas Main S:, -.-.;. ~ ~ .... :~ ......... 1 Gn l 0 : I I I I I I I I I ,___ I I ....... m.............. Hn ,, ................ : I I a I I I I I I I I I I I \ I I I I I I I I I I I I I I I I I I C ,--c,,, _.,. I I I 1 Pref I --> G ....... ........ : :If--< ~p In ectarl(t : p ; REACTOR PROCESS ORIGINAL CONTR O L S Y ST Er 1 --. -RE-VISED CONTR O L S YSTE M I I I I CONDEI-JSER PRODUCT COMPRESSORS SCRUBBER Figure l

PAGE 10

, 7 2. 4. 5. CHEMICAL EBGlNEERING EDUCATION Dec. 1962 reactor pressure necessary that the compressor suction change in insure the appropriate flow through the system. Some means of compensating for very sudden changes Since it is desirable that the be constant, it is then the manner required to in reactor pressure mu s t be made. To account for relatively slow changes in the system, such as fouling in the effluent lines and gas processing apparatus, some means must be provided to adjust the compressor suction pressure to compensate. Because the gas handling system was rather distributed, an appreciable delay time or transportation lag could be expected. A satisfactory, It was also control system existed on the process but was not entirely especially when sudden interruptions in energy supply occurred. quite sensitive to partial plugging in the lines. For a detailed study was warranted. The work began with these reasons and others, investiga' an experimental tion. installed primary transducers. Twelve channels of recording oscillographs were ments being made with very sensitive and responsive pressure drops were sensed with Statham strain-gage transducers ments with multi-turn potentiometers, temperatures by bare measure Pressures and valve displace thermocouples and energy by a suitable combination of variables. Figure during the l indicates the experimental process work, and scheme, the the original location of and revised measuring elements control systems. The system 1. 2. following tests were on open loop or manual conducted control. at various levels of production Dynamic tests to determine and other pressures to the response of reactor, compressor a. b. c. d. input, change in reactor energy change in displacement of change in displacement direct injection of an existing compressor bypass of existing series throttling inert gas into the reactor. Static tests to determine valve, valve, a. b. performance characteristics of valves, sensitivity of reactor pressure to valve displacements, with the suction, 3. in the influence c. pressures and pressure gradients at various locations Extended observations to note general process behavior, system. of dis turbances, and performance of control system. .c::::==:.-=~== -=---r...:; ... ___ :::.= -:::=:= -----... -----.. c:::::::c = ~~=i~~ii.smr:!51-~=====-c:ccz: ==-=t::."i.. .t:::: .. -=c::.:r:::::=::st:=: ~~ig~g~le ~== ~ 1 : ::: ..... t: --_ .. ... .. -i5 .. 1 -~1~ Ji-: = n, -:,: : -:: -u:::t ::i.:; ~:t: ... !,':-:'.-:-:, ... .. -: n rE i ii= :,:: .:i ..... ..::-: :r ,:-1 ~. .. -. N-:.; :: ... -; 1;-, .:--=:r,o, ...... ... ..... ,.... ~ = ::.c: ... -' ...... = ~1 -. .... ~n"1: 1t:a1 .. -rt: : : ~. -=1 .... .. ~.. ,... ... -~. ; .:r, :; ~ ; ..:r. "'! l "' . : : :! . ., .. r:: C..:! : i:! .. .. ... 1 : .. I:; : r: ~ti:; J'i~t-ti!,l _,,. ot .... ... : .-, ; .. .. .. -. ... .. .. 1 .. ... I"' ... : : 1 .. .. . ._ :l: .. ... .. .-, .. I : : : .... .. .... r r.: : t i.t. . . .. I o ...: : :t : :._;1: .. .. ..... H .. :~ i, :.. i ~ ,.: :!.! : : .::: -~ ... . .. .... ,.!.!.!;.!.: ..! ..... .. t ::. ::; ; ::: : : : ::.: ,.. : : 1 : :: 1 : .. .: ; :; :,_; :: ...,c c s : J.: ;S : > : ... ~ :., c t c : c +cc ; .., J t : : : ... ,-; 1--_ .. .. ,, .. ~ ,,: .: ~ :: :: : : : '' :. : .. ... : .... .. ... t : :: ::: :, ..:. : !. ; :.~a :: r: : { :,: : '.. ... ., I ..... -... .. .. I .. ... .. .. : : ::: ... ; ~' I: ; .::: '; ,::: ,,: ., ., .. .. ... ... ,' .1-a ,,_._-';.:.;_ 1-, .,.:: +c cc 4_ c:_ ::!. ~ :":.! 0 :;a ,.; 1--,s i-" .. 1 : ; :::. ;:i: : '!! .. ,; :;. ; ; .. : .;: -:i :: ::L t: :: : :.: : ... !' ": I : : :: t ; .. ._. 'I . .. ;; ~ ~ : ~ : : : i ~ .. :" I: : ~ _; 1..!. : t. .. : -~ : 1 .:r .. : .. .. 1 : "" "t ... '' .. ; j l...:.:. : : f. c;: "' .. l ; .. : "1 . .. .. .... .. . . .. ... : ~: : :.. :.. __ ) H : .. I : 1 ;. .: :..-..!.~~ t .. '" :: : : .;:, : J ." :.:: :d: ...... ... ... ... t:;; _.., ___ ,.,... ~ .... .... ..."!' _.,. -. '-~-; 1 ~~~ 1 =:;:; .~ ::1 : :~:'.: ; ~:~: ; : Y=: ~ ~ -. ": ~ _~= :" '' : : : : .. :.:.r ::.l: .. ... .._ : .. .. .. ... ,._ .. -l. .... ". -, ':J .... ~ r. L.,.. r.:.r ~l -:'"; ~ . ~ ... .+ -... ~ -~ ( < 4 1' ....... !: .. j : : : ... : : 1-: : I. .. .. . . .. ::t: .: .. .. ... ; ;; : :i = 77. ... :~ : : .. .. .. .. .. --: .. . .. .. ...;._ .. .. : :~: 1 ... "' t . ; ,~ : ;::;-' -.1:l ::t::t ,. "' !:tt. ,. .. ::::: ,,_ c ; :.: : ...... ... 1 ,. . "J ... -,l :. t:t! :1:1 :::-: t :: 1 .,. '1 .. .. :t 1 ; ;.: t I ; t: r:: ~ I ... ... .... J::-,. .,_ : : I '.' ... :.:':; t. : ~ ... : ~:; :u: ... .. . . 'j I l ... . 1 : : ;. .. -. .. ... .. ;:!, .I I : L L,. 1 .. . .... ' .. I ? : ' ... .... ... I ; : : :. : 1 .. .... .. :: :::: : : i: ...: ~s c -1-. I .. .. .. ... 1p: .! 1: .L : : :: ; : : ,_, ; :s .. ... :.!.: : : . .. . .. . ... !I .. o .. .. ...... ... ..... -'+' -"C'f'--'-'l ~ : -: ~-, : ~.: :t .. ; .. .. ... .. .. .. ... ... .. .. :..: .. 1 : ... .. .. ... : .,: : 1 .. : :: : ; ~ .; ._ .. : : 1::. ..: ::: ; : a ..:...:..c: ::.+ ...: ~1--.i.:.:. ~ .o .1: : : ~...: ..: ...; :!-I ; ..... . .. .. .. ,.. .. I .. :. : ;-..::. .. I .,. ..; -;. l 1 '' ., ;.:; ... 1i'' .. .. ,L .. \ 11~' .:?~ :;: 1;: r .. .. ,, l -, f1 !: J 1 1" : : .:: : 1 z )tt ,r :.: : I .. : : .. .. ;.; :i: .. :: : .:: :.:: : : : j : :: .:.: .. -: -~ ~ .. ; ~ : :r J :: :.. :! : .! : 1-" '-'' : i-:' :: ..: s i-:a ; .. ; : : j... .:.. .i.: : : ; ..,; a: ~ :c -1-.:::::,:: ..:.-1 ..:. :: : ~ i.: 0 :c c ~ ; ,r-_ =. : ~.~: : :;: ~;;: ;; .. ;~:i~t ~\ ~ 1: s:: : ."= .s: ": -l-' : : ~ : ; : i .. ; : :=r t ,;~ ; : -:: ::: ;. ~:: .. ---1 = =-r;: -... = 5ii r ~:: !..ll = ... .... ... ~ ~ .. ... 1J1~ .~ ::.n ... ,!"?".: ,. 1 ,:.;: .! !.J .:: ,,_ ... ;, 1. -.. 1.: 1 i::.: i3 -~ ::.c E .. ~. _.;:_:,:.. .. .. ... ... ...,._ ----1 ......... !l..1 -.. ,. ::r. ;::. :;-:.: .... ....... ... --.... ;,: .. .. .. i ri: .. ... =.-:; ;:-r t"... .__ ...... .. .... .... -.. -. ... = .;;; : ir;; n ~ ,:.:: .. ;JJ ,. ,. -:.-.:..1' L:":' -j; ~ : :rr 11. I :j :~, ,t U. :::! ... ... .. -. ~ J I .... .. .. .. :: t:.: .. .. ... ... ... ..,.. ..... ---~~ ... .. --, . I" :r.: :.r: ...... I~ .,..,, r 1........-: .. -:-:-:+ r,; :!'.'.: ~ __. _,,. 1 .. .-:.-:.: i::.: ;r ~ -: .: .:.:: ::;' ,!:"'I -.. .. r = r-: ~ .... i ;,,. :J:: t ;_".'.:: ~ .. ... .. !.: :; ; -' ..: s.i.: ..: c .i; .::..:i 1-s c ~ ; .-: t .. . r.-:, :%1! ;Jr .. r. 'a ... +::]:-~:r,," :;;; 'l-;j l ':.J !' i; : it t" .. .. {I : n:.. i :. l :., ;;: r :: t ;.. : : : ::: ~ :: = = ~. . : ::-.. :: ::. .I;;:;;, ..., ;'ti-:.;:.; "1 .. ::l ... ,Jc t: ;,t" ,:t:.:: !l.1'1 ..r ; -;:::., :.!':.! ... ::.u -~ ~r ;;.-~ .. 'l!. -_,, ..... I ..:n .::: .. If l .... -, .. d .._, ,, : ;;;..:::.11 .. .. ,,_ "f ,c t.~ t .:::: ..-,. .. I 1 : .. .. ri.. ,n,. l"tt" : u ,, r ::r:-r i~: ... ... '. .. ~ 1 : t .. .. ..., : r-i' .. .. .. .. :..;.:;-: .. ,. ,: . .. -. ;.. ... ... .. + I I ; ~ . _. .~ .1:_:, 1 ~ -... -. .... ~ ::t .. : r .. :.:. ; ;,. .i..: "':.' = .... ;-:,_ --=~ .'1. .. ._ .. + .. ... -... :.::. ... ,. -.. ... .. [' ... :-: ... ... ~= .. rF.: t,.. ... ...,, .. .,, l ;Jj; J! I j t n; ,. 1. .. li'I : :.t ,IT;: t ... ;j : .. .. .... . r" 'I: t I .. . .. .. 't.:. l : .. -.' I _, ., .. I : tr 1 ., : ... : .; :.: :i:. l ;iJ ... .. ... .. ~. ,., .. :! I l I ll t : : ~ 'll; ,i = r...,;. .... .. .. 1 .. r' .. !r-..: ,:~ .,.,; h :;ic ; 1 ,c n !--is u11 "I .... .... !' __ :; : ~;,~:i::.: r;: l: 1 1 r 1' 1, .... I 1 ~ 1 .. .. ; ... ~" ;J:p. ,., ... :r-t I ." l : ... .. I .. :i, i:. ;:; l t ,.., 11 111: -::: : ~: t... :,.: ,;_ I J f : -r::. I' :1.1 .. l : r ;n i ; l ;_. 1;.; .; '. I ,., : ... ... 1:;l t r: 1-'.j ~ ... l .. .:...1. ~,r. .... .. .... .. .. .. .. ; 1 ... '! : : . .... .. . : 1.' : ar j!-jj ,., .... tn .. .. .. -11 l_ ti-' .., . 1J I --,~ 1f" : ,... ~l r .. :;r-: u I:" : ll .. I ,. .: t; : ; ::. !f ,.. !t _,_ ., ., _, .. ,. lr. :: : ,.t;: . .. : r i .. r: . ... ol I .. f: .. t~J.: ,1 ... .. .... .. .. I ;;.,; .. .. .. .. : : 'r ; ii! I, : I' .. . : : t. ... .. 1 ff : t. l ... : ,:i 1~ : : .... .. I I: : ., ... ... '.' t ... t : : : :.; l .. J .. :: : : !l .... :u ... ... ::;. .. .. .. .... ., I ,:., ;'. 1 ... .. l :.: ; .. .. l .. "' ... .... II' : : '1 : .... .. .. .. : : i: .. '. ; : I" ','. j: : 1 I I" ... .. .... i"" .,t: 1; 1 : ; !! n:t .. .. .. ... ... .... .... .. .. .. .. .. ... ... .. ... ... .. .. .. .. '. .. ... ::: .. .. ,: :: t :.:..: ; 1 ?. :.!: :z ,.,,.~ t : .,. : ... ,, ;.:..; ; ;~ ., ., : .. ... :-:-,:.. .. .. .. ~. : l-' = : ,4: :: 0 1 ":.: : +: .,.:c+c ;:~,; ,:::.:; t.:.:..:. : f ,..:..:..:::..:.i.: s::.:: c +cc ::;c c c 0 4-: :t 7 :.: --1 .. .. ... .. .. .. -... --1 1 I :-"'. == :~ :1: : ::: i: 1 i :: : :;; ; .:~:::tr! :-= -;-: :,: ~ : ::.;: .: ::: r ..:: .. : 1 ;::: --1 .. .. .. .. .... .... I il'(~,l: :,;.f: f ::-: -: :-:,: _;~ 1 .. ,, ~ !' :: ; ~ ,: .. : :-:7:':-;;;: ,. 'l .. .,_ ... :: ., .. .. ..., ., ,. ..:.: .. . .. .. . .... ... 1 ,. t.. t. ,. i, c.c -i. -i","" "+".'-+ '+' !- + .. .. ... I .. .... -. l H --.. .. :: : .: 1 r: : ; :.i.: ~.:! : ... t.;-: .:" : .. : ~:._r = :. : ... ; .. ... :;-: .:.-;..-: :::: !::, .. .... I., '" : r I~ .. .. ... ..... '. .. -. .... .. -1 ..... ::. I ........... :: ~;.l: :: : ; : ~ .. ... +:.:-+"-' +--+ .. .. ... .. .. . .. r I .. : :;:_! ... .:!:: :;... t ;::.:-... .. .. _.._. ,: '.:'." -+;;.; : ... .; J-:. :~ :.: :H .~:: : :E :...J: .: : :~: t::: :: :, : ::.: :.; J :_:;: :':i : ~ : ... : ..: : : : : : -~: ... + ,+..-:. a s -+s '-+ -+ c ','-1-'' t,-c i-:s _; i--" + ..c c + -:c c +,.; ;;., .. .. .. 1, ,. t : 1 :: 1::, .. .. .. ~ ..... .. :., .. :: ... : '. :" .t: .... .. : :::. !' t: : _, :~ :. '! : .. . .. ... ,, ... .. .. . . .. 1 c .__;:. .. .. _.,, .... .. .......... ... .. .. ... .. .. ,. I : !, ~ : : r: .' ... .. .. _,, : n :; . r ;::: '1 .. ;-: : .. t ... .. r. ':. .. .. .. .. . I .. .. : ... : : .. .. :! : --::.. I ....... + ... ... ... .. .. ,._, .. i. ~ r: : .... n:1 .. 1: :ri; : ,: ;r : ":l U .. .. . ,_, ~: j:J ~ :-_,, ... ,. ': rt : :1,: : r: :~a:..::: .. .. ; .. .. .. .. .. -~ :-: ,. .... ... . ., 1 ... : ,: : ; ,.: . ..... . j .. ,. : J* ..... t .:. ...... :ii! :::: : t" : : f '. : .c.. .... _, _ :.. :' "i : l. . : ... t .. 1 .... ., .: ,;; ;: ... I .. ...... I 't:. r: '1 ,. . . .. . . ::1,._" .. 1; ," .. . ; : I : .. 1 l" ". ' i. : : ... ; i I : .. : .. .. ...... --.. : ... 1 I" '. . .. .. .. :r.~ ,: l i ''i .. :-:..'. '1 . .. ,J;1:; : ~ : ~:~: ~::: = : :-: ~ : :: ; _;, .: .;:= ::r; ::?!. :: : ::: i:: l! -'' ~ : ;_..:.;J-: c'~c':f.' C c:'f -;;,; :' ,J- ,'-' +; : 0 7 t: ;-;t: : :-r ', c'ir c: c + c c ,tc : 7 ; : f" ;= r ,, : 1 : .. .. . .... : :: .. .. .. ...... ~ -=L. .. .,,_ j: ,:... ::, :~: : :. :: ~ ,: :..,L.: ,,:_: --:; ;:: -. ;: -:_: _;:..i: oh ,,. .. :~ : : :: .. ~::: : ::: r-: --::: .~: : .. 1 .. .. ...... ... ,._ '!t t: .. j"! .. : .,. : ;.. i; : :i .. I .. ' . : :.; ... I .. '. -. f .:. .. ... .. .. I . .. .. .. .... 1 >!'" J .. ..... : :: ~ .. : f: :,,, ,:_;~ ,. H I '.I ,. 1 . --. .. J.'' :r:.:: .. [".J. : .! ... .. .. Ii: '. j : t 4 ........ ~ 1 ... : 1:..: .... I"' .. .. .. . ..... .... ..... 1 ~: ; ; :... ; ; __.. I .... --. .... ... .. .. .. .. l::. -; ;::. "1 .::, : .:.:=. -, . .. .. .. .. :t : :. ..... ;.: .... ,,._ .... -.... i= .,,__ ~::-..... .,.; _; ..... ~ -.:. i....-:-: --..:.. ...... ..,.. .-:: ... .. __:--..:.t..-:'" .... ..... --.~-i-t= ..... .. :_. -' .. ~ .. .. ::. ;...;: : :::: ::.:.: .. -::.:_ :.:r;.~..: .... .:.:: L..--=--: r--.. ,- .i..;.. r.:: r '. T ';!, ,: t ~ t:. ., .. ::. : Ji~' p ,: i' .. :: ~ ::t_: ,. z : t ~! .:~t :--: :: 1.::: ::: ;:;l r~ .. ... .. tr:1. 1 :.:. ; i,:'.:-1-: .: : '"'. !: -+: '_ 0 ~::::-!':,'.c:c:+ c c~c' :+1 : : :: : : i., : ; 7 : 7 le .~:,: ,: 1 0'0 : s':r'a -c : + : ..c e c ;:: c -1 :c::.:. f :; 1 s.s: 1. :::: : l! !: ::: : i.: ,:; .: +: :: ~: : : : J h 1 ~ ;_ c ::_::, e c~ : ::'f ::c ;:. : : .. .. -; J l :, :1-; .. l ;..:,..... 1 ... ,_; : .. : ; ::: ; ~ : : :... : .. ,..:..! : :..:1 .. ; 1 ;: ; :1.:: :;:t.:r .: : : .. . '" ....... .. ... .. .. .. ~.. .. _,.. .. : =~ -! ;:... .... -.. ........ 4:-:~ ;:: :: .. ......... i -" .. .. ='?.":. ::;.: : ,;:; ..... .. -~ : !~! t; ii~ j";.\ t-. ~ ~t. : .: :~:: .: ; j' .. ;;: .:: :. .: ...o tr. t :-: n :" : :. : :. : ~r:! : :: ... ... .. ::. i, 0 ; .;;:,.ic.." ",f'--'-.+'--1, '-", I '" .'i. 1 ;., i, -: c-t ,.,.... .. ,.-., ', .. :~:: 1 .::;; r:: ~i ~ :;1~ i=; -:~: ,1 :1~: ;ll 1~ : : :: .. ,:; ,. .. .... .. .. H, .. ... :...:. !..-= I' 1 : 1 .1 ; ;: .. .. ..... .. ': ,i1 ,::-"! ....... .. . .:.:. ~. f: ; s -i-; c_ +-'1" .._ ~h c ; .':.L '--1'-. :' : ..;.:-" ,. _; i .i i t;: : :" : : .::: ': :1n : r.: T.: :. : t; 1: : :: ; t.-:. ::-.: n-; 1. 1: :: : ~ t: ... .. ... .., ~ ..,f... .. '1:' ~; ;;,;_ .. ,, .,_ -P --.... .. ,.: ,. ~-. .. ...,. --. . ...... ----' ::.:: : ,_ a .. :; ,,_ L: t ... ....... .. .,. _,.. __ .... ~:: ..:: .... -. :r~ .::ri .... ..., !1 .::: .. ::;:,.. ....... _-:;'.:' :"tJ -:; ).;1 :..: ;._ .. :;.:. "*.;:!. .. c~ c ..;. :.,, : 1 j.:: : !. ::i : 1 ~ !. ,:.: -1...: +c c ::i '~ '~;-: ~-~ __, r rr! .... .. r r'"': -:-r~ .... .,.. :, : :--:r. r;n :..:~ j !.: . ... .. ..... ..., u,, "".: .. ;,: u ~ ; :1 nn ~ .. :: ~ 1.. 1 1 ~ t:. ... !::~ :rt: ; .... ,:t .. .., ...... 1 :: :.,_ 1 ., .... l r ..... .. S'.1, ,!. H U ;:t!, ~ 1 ; c : .o ~..: ;: -i-; : -'-, -'1. "' ~ :::i-:-;.. .1_ :.;-i-; ~ :.: -1: : .,; a: ;::; +s 1 c o l, ,:;:;;:r ::. ~ ~ =:;i::.;: ;;: 1 :,=1 ; ,... .. .;: :-::: ... ::_: --::: ia: .. :..-:1 .; l ~_r-:.:~J ... .:..~r- =.t! -:: 1: ... : 1 : .. : .:.:"!I 7: ~ !; ::; ; :li, r ; : -:-:: r;:-!. lTu :r:,_r n: r.h ... -:s; L:.. : tr. .. l .. .. -~ -. ;,:.: '-::'";. ';7,; --~--=.... --: -~.:...-;-::.::.... -n ... .. .... ... ...'. = .. -. ....... _';:: -.:r. ..:;. ,:...,. .. ,. l ,:.:: ;1:r;:--n !.d ni : : -: ........ .. .... .. ,-.--f" .. ; ... .... ..... ,.._ : Tt :. .. ... -~ . H: : ,;t ,_ : .. S; : .5 1 ,,1l .. .... .. . ---t ,. .., .... f,' :.; l; .. ... .. ... : :: "' :~ 0! ;, :: :. 1. : ' .:.:.l! ... ..::..:.. ... r. : ,-l .. .... I .:...: ,. .. ,, .. .I -, -. ::. 1 ,; .. : '. = ------r --= 7!:-.:t.:!. == f t .... r:.. .. ... -...... .. ... !T'"!..,. l. t::: ; ... .. --... .... rt:.:!";! ...... .. .. -,........... 1 ,-,., ~ n ;;r :..~~~n,:: .!l 1 l' ... .::-.: ... = J~ --:"::. .. --. .... -----i..- t '1' ,!.! .:.r .. .::. r _,:. ... =. -1 .-:..4 .t....:-.J. ; .. ::: l:C. -I = 1 .. .. ,''2.;" .. ,, 1 .. l :1 ;., ;! ... ' ; ,l :.:;: I : ... ... __ ::' : f '. .. ,. ., r ''" l 1~ :: ,; l *...., !' 1,;. _, .. .. '. TIME '-4 I ::.: ..... r .... .. : ,t"" .. .. ... .. ;t 25 '. : r:" .. ~-~. .. .~ ...... .. ... t .. :.I :: : . ... : l. :. ... .. .. .. -""1 '7!" :t! -: ~ r.:. !~ :.: .., 1 ;.. ,,,, .; 1 .. :: : i; ... .. '. ,_, ... .. ti:: ....IC' ...... r. .. ... ,,. ,_,j, ':":""..: .... .. t "l ... .: ::~ -:,;. ..:~ -;;._tl -:.u: D IVI SIONS =ONE SECOND Figure 2 !I;~ ... : : .. .... .. -: ; ... : .. .; ;.. ..:; t -!i;'! f!" ~..... """ 1""... ., ,,, []~ __.Hr+j,h _,, .. ... ... ..., .. ,....J r.. -, .. ,. ~ -,..._. -.. .... .. .... : ... l ;.:..:_ : :'" .. :.;;r ::;_ ...., .. .. :-7.t ~.!1 . .:-. !.::: ....... ( 1.4 0.4 -0.6 85 80 75 15 o N :z: N :z: (/) :z: H C!J H (/) Cl. ::, (.)

PAGE 11

Dec. 1962 CHEMICAL ElfOlNEERING EDUCA~IOW 8 4 o Et-f:=1f-=E~ r 3. -3-3 1 3EE.-E :.-E: 1 i ~-~ 3::j=.=_:~ -~~ E := = =.E: = =-= = I ~--~ .. ? .. 7 -J 7 ~ ~ ~ -~ T= ~ ~ ~ -r :-7:' ~ 7"-:-:-:-::-:=:-:-r-:-=:::::;::-:;: + 4 ::::::~ H-~ ::::::t=t r ---, ~----i-. f h --. ---~..., ~-.... "!'"' ---~--~-.--__.,.. -i ,~ t.t..i, 1 _++ -+-l:_.+_.""t-i:----t---\ -+-'r-i-'--l -;-,....,.-t --,.. .. ,,.. ... ... -~ -. .. .. .... ... ... -.. 1 ~ .._ .. ,_~.;.._ i 1 ~ -1----t-+--1--f 1 ,, ----... -.. r -t -l--+---<--l --1-----' ---,,,..-+-------_ _. .. .. --.. .., .,;,,_ ..... .,._ = r~ ,~i-1--_+ _._ ++-+-+--+ -I I ~ r-t-'l-l--1 --1 0 .i -. .. .. ~ --~1 ... ..... .. _,. __;_~ ~-' .. e 4 5 d /'\ p ; :::;;=. : ::. .:. :.:.: -.:::.:. ~ ~ : ::.,-=,__ :: ~ ::_ ~-= ~ ~ =; --l---~: ---i~-::.-::. ::J _l -.. -. ,--,--:--1 ~-1--...; .,.._... ..! -----+-,~ ~:t:;-::_ ---~ -;-t-:-__ ,,_ -t~_ -:,__;_ ~ -+-~ c_ _,;_ .. ~ -: _, 1~-h-~ 1-=i:~~ ~l::J l --i--/-t--l-f--f---f--!--+ --f-+f--. ,-+-+-,,-4'. + : _. .., .,..!.. --~-r, -----.I I I j: I . -, 1 -r-r-i---+--l ~ t--, 1 ""-'. +j -+1 -+-+-+--"r-~+-+, ...1-,.-+: --t -+--:--;I ; l f + +! 1 1 ---t1 ~i -i-+-1--+-i l .. -1-1---r- rT t -t-t-t-"'i-+-t-t--t-H--+-++-t--t--t-H Hl-+--;1 ..;-... 4-\--l-l-~ -t--4' --1,....;-..; a L L ~ ( 7 1 i r l 1 1 I T i I -.I l--t--J'--i -l--i---.---;-+--+-1 I I I I I ----+-!--~FLO W = 2,2 l -ri"T-H"Ti-1-H-t-t--+-t-t-t-t-H-t++-+-H-HH-H-+-4---4-+~ --t--1-: ~d ec.. l1 H-t-t-++--1--+-+-,,-I ;1;:1t"'t"t".t.tj."_t,_ +_ -l if.-++ f=: =: 1 -_ 1-+-I l-+-t-+-l'-1-t-t--t---t-+-t--r-t-t-,1-1--t-+-+-+-l-+-l-1~~-= -+--.I 1 :f-11 ~,_, i-, l i--+--i --1 i--r-t-t--r-t---t-t--t-t-t--l-+-f-+l-l-+-+-~--l-+-l--4-1-+--t-1 '-l O '::i : t _; =_ -: =::,-=:_ ~ --= .::'. :-='r.=-, ,,-:: : .,;; ,-.:. ,,._, ~"" .::r = -.. -:-; _,, ,._ 1 ~:~ -1~ 1 ir ,.. Fr ; .,. .. .... O ~ ...~ ~;;_;: ':.: .;.. ..- tN-,O o~ :~ .. .. ~ ~ ..,,_ .. 1 ... ,.. -=-.t::=: -=--... _:.::::::;::_;:J.;:;--= _:'?' m = =;:-2::i -=""--~ ::-= ..;_;; = ~ ;,.::~=: ... -: =: :-. .. : :~ .. .. ~ 71 .... ;..-:: --- ._ ,_, ~ -;r :.: h ~ J::"" ,., -~'"':t .: :..&..~ ,.,. .: ,.,, .. 1 .. 1! _, I j _; :e:= -... .:--= .. -.., ..... ... ... ...::. ~ ~ = =' ::. t :".! ..... -. ~ t .. .. == ... -= ::= "" -.. .. ... ,. ~ ~ F."' -:.:::,_ ... .. -. -, == .. .. ~ ----l= ; .. 1 .. .. ..., =.. ::::,_.r ---,--;:..; / ,. ,.. -_,. -.. j> -., .. .. ..... 8 = ::, -;.:; =t= s :::;:: -; .. .::: ., :." =.r.; ,, ; : j ~ I : J ,' I:;_ ::: =i= = ~ -.. -;;; .. ~. ~. j ,~ ~ J --:.:. = ==== ., ~ :r .. .. '... .. ... ... --.... -.. ,...... ., __ ,1,., ,.., --' -~ 1 6 420 t: ;;,;: ... .::: :-;~ ":'= -... :, _. : H :::. .. : : 1:. .. 1 0 ... I ; 1 l .. ,1 : :. ~ : -: = ... ...: __;_ --~ ~=I 25 30 .. .. ..... -, .... l ~ -.. .. : .::~-~r --J .. =: t ::... ::: ... _ ... -_ -: --;:;J::. 1 : :... : 1 ., 1 r : ::: 7 35 -I' -' "" E-~ I f ., .. f .. .. .. .. _. -.. .._ .... r. ::i .. ... i 1 :-t ..::. -. . . t ...,_. .. ~ ,: ._ : .. ;.. :-; .. r:__ -: .. :.. -,: ::: ,-: ~ : ~ : : -~ J --~ J : _t .; --. ---. .. "' --t-f -- .... ... .. .. -.. -;-...... --+.... ,. --r,. ... .. .. --:, : -.. ..... -4 _._ ,, .. .. ... :j ... .. r .. .. :::.: I"-.... .. .._ ,.-,._ ..... -~ -.... .. ,_ "' 1--. ~ ,.. ., '__, --. .. 1----, 1 .. .,._ .. _., 1-.. -r ,_ ... .... .. 3 .. --_J_ .,, _, ---.,_ .. ..... 1 ._ .. ,.. _.,. .. ., ,." I 1-. .. !, .. .__ -, -~ r ,.. _. ., .. '"'h t -:-:t::::i:-:---1::r.: '-" 0 0 .. __ ... t-P" ~ --... -----45 ---"t _.. -~ __ __ ,, .,.. --f.-.. --50 __ ... ... . ,,._ ,. t 1-...... ,,, .. .. -55 VALVE DISPLACEMENT, d, % OPEN Figure 3 All dynamic tests were conducted by the pulse method ( 2 ,3,4) and were reduced to frequency response via a digital computing routine, delay time between records being extracted visually prior to processing. In this method an input is caus~d to vary in a pulse-like manner and the resulting outputs measured. In Figure 2 are presented reeords of typical results from a test wherein nitrogen was in jected from a presstl!ized vessel into the reactor in order to measure the pres sure response. Figure 6 shows the corresponding frequency response information Important static data were also obtained, typified by Figures~ and 4. In view of the nature of the system, the requirements, and the information obtained, a control scheme was formulated. . The scheme visualized relatively fast control systems at the terminals or the process with a means of loosely connecting the two. l. 2. 3. 4. A tight controi or reactor pressure by regulation of the directly in jected gas stream. A tight control of compressor suction through bypass control. Adjustment of compressor suction pressure in accordance with energy in put using teedforward control or the set point of the suction pressure regulator. A meana or recognizing slow changes in system behavior (fouling, etc.) b~ aenaing reactor pressure and allowing this to adjust the set point or the compreaaor auction pressure controller, A complete block diagram repreaenting process and control syate~s is shown in Figure 5. S,mbole are defined in the list of nomenclature The objective or the experimental work was to determine the character or each block. Upon complet~ng the data reduction, 1n1'ormat1on waa on hand enablir.e; a detailed dee1gn to be made. _; S inoe botb. the gaa 1njeot1on ~ontrol and terminal b7Paa1 control ayatema ) muat be taat compared wi tb. the other control 1'unct1ona each waa oona1dered aepara.telJ. For compressor auc.ti<:>n control the t'ollow1ng criteria were used 1

PAGE 12

9 0 -10 -8 -6 -2 CHEMICAL EIGIIEERIKG EDUCATION .. . .. .. I I .. I . ,i ,:-..J ~ L ., tw t .. .. _, I j ,. f \. \. l.L i 1 .. .. ........ t -t l .. .. I .. I .. I I .. .. f .,. ,; .. ...... . .... .. ... ... ... 1,, ..... .. ., ,. ,, ,1 ; ,,.,# ., 1 l;,. l t l 0 1 E (" >t ~1 .. ,, 1 .. .. .. a. --. .. .. .. ... .. . .,. .... . -, r ~ # 1, i I l "' I I O j . p.,. o I . r .,, .. ~ ... .. .. ., ,, > 1 .. ... ,., .+,., i ,, I I ~,. \, . .. .. ... . . . . . .. . .. .. ~ .... ... ..... ,,_,, I, + t .. l I I .. I ,._ 1 I ,. ,._ .. ,\ . .. .. . t I 0 h I ', .. .. 1 lo,, \ : t I I "I .. .. i ' .. .... .... .. .. t.. .. ... ,1 .... .. ", .. .... I I I o I I \ I I o .. t + I .. .. I ,,1 ... ... ,. > ._ .. ~ . ' \ ,.. I \ .. .. I ..... ........ ~. -, ;~ .. ; -' .... .. '. .. I t .. .. t ... ... . . .. . ,;,.~, ~ l .. 1 .. -, .. ..... .. ... .. ... .. .. ., I .. .. .. .. .. .. 9' h l ... .. . .... ... . .. .... ..-, .,._ .., -.. ''9 ; ... .. 1 ,. .. ..,. ... .. ., ..... ... .. .. .,._ .. .. '" '" u ... ... ( .. 1 .. .... .. 1" '~ .... ... C I .. .. ... -,_ .. I ., ,. __ I .. ... -" ,.,, . ... -. ... .. . ... .. .. ...... .. .... .. .. --' .. .. .. 1 '. ... .. .. .. _,~ ...... .. ........ ) .f ,. .. -:-- 1 I ..., ~ .. .. .. __ .... ....,.. ... JJ J .... ......... .. ,~ ..... .. 1 .. .. . .. I .. .. ....... ; ,,. ,J .. I .. . I j I f .,.. I .. ,. .. ..... ... I .. ..... I 1., I I o .... .. I t -. ... .. .. ,, .. -1 .. 1 ... .. .. .. ... .. .. .. .. .. .. . .. .. ..... -to I .... .. .. 0 0 M . l J ., .. ... .. .. ... . .. ". ..... .. 1 -, .. .. .... ... .. ....... ... ....... .. .. .. .. .. .. .... .. .. l. ,1 .. -J ,. ,, ~ >- .... .. .... . .._ .. 4 ~ .... ... . ,. ........ ~. -. ... .. .. L .. . . . I t t of t I I ._, _.., _._ ,, ... ,, .. .. . .. .. ,. .. .. .. t I "" .. .. .. . ... .. .. --.. .. ... ... .. .... ...... .. ... .. ........ .. 1 ....... .. t .. .. ..... .. .. .. .. ....... ,. ,, --~ ,. ~. ..... .. ... I ,. ., -.. ... .. .. .. ... .. ... ,. l:: :." :: .. ... . I I t t _, . . . ... .. ... .... .... _. ,. t' .. .. ..,.~ ., .. ... .. ... ... .. .. .. -i .. ........ ... ..... .. 9i ,. ,, ... . .. .... .... ... .. . .. ;. ... .. ... ... . ...... ~, ... .... . .. '" : ;.: ... : ::: ; ; ~ .. ; :. .. :. ..... ... ,. ... .. ......... rz ~ .... \ ., ... .. ... : .. ... . : = : : ; : "" ... : : .:; : :: ~. . .. r .. .. ..... ... r .. .. '. ,... I f I .. .. O, .... .. ... .. H ;, ~ ...... ..... i,. ,,.. __, .,..,...... t n ~ ,. ... ... .. '". .. ... .. .... UN, ...... . . . .. .. ,. .. .. .. I -! -, .. .... . ... I ..... I : : :: : ~ : :.-: : .... L .. :: ~-. :, : : : : : : ., :~ :: :::-i ~ ~ \, : _:. ~: .. t .. ... ....... ... .. . :" -... _, .. .. .. ... ..... -. .. .. ... . . . .. . .... .. ''I ~ ... i .. '+o" I . ,. .. .... .... .. .. .. ,, .. .. ... .. .. . . .. .. .. .. ; .. .. .. .. ... .. .. j .. .. I I ... .. .. .. ... .. .. ... .. . ~ -. .. .... ... ... > .. . .. . .. .. . .. . -. . ..... . . . ... .. ... . .. . .. .. ~. .. . . .... .. . . o I .. . ' . . . . .. . .. . . I . . ' . I I ' . . '. . . . .. .. . . . { t . ' I .. ... . . I . . ; { .. I '-. l J I I .. ... .. ,. .. ..... . : 71 ; : ~ .. "L .. .. .. ... . . .. . . .. .. . lo .. 71 / . . .. , . . Dec. / .. ?: .. .. j ,. I I . . . . ---------i----~ 1962 .... 1 --,. .. .. i -~( ~ -: ... f .. .. ,_ .. ., .. .. -: """ -r ".... ,_ ... .. .. .. .. JL ~ -~ .. _.; ,, j I ... t1, .. .. .. ,J .. ... .. ,..'r_,.., .. .. I .. t ,-. i . I I .... .. ' 1 .... l -+-~ ~ --+---i--r --.... .. ._ ....... .-.t~ ,. ... 1 ....... --1 : ,,,,; : -" ;, ... : ... I __ ,. .. .. r -: ... ... .. / ... .. ... .. .. ... .. ... r -: / . ._ I '. .. I t I 1... ~.,...4 -e l -H : i I_: I .. .. .. l ,. ; .. ... .. ~ ,, .,; ,.... .... I ,._ -1: ., .... .. -,-. ~ .. .. ,: ..i-L. ~ .. ... .. .. .., : I : > .. .. I : I I t l l I I ... I I I i . I I .... . 1 10 20 40 60 EfJERGY .,... L a regulator system l e disturbance-suppressor and assuming in control valve displacement Ae a variation distrubances were 0 D Similarly for the gas injection system l ~ref 1 0 For the suction control system both criteria are satisfied e maintain the To usually requires l since then l I required high gain in spite or that Gn contain compensating attenuation in dynamics. Thus 80 1 00 equi vs.lent to if and/or Gnn a

PAGE 13

Dec. 1962 CHEMICAL ENGINEERING EDUCATION Co ntrol System No 1 I I Hn I I 1 I P' n I I dv I E Gw .... x e + G n Gn v Gnn I I l I I I ................................................. ............ p n, Gpn Ji t ; -, GPP Gap c;n l r I Pf pf, I '"" + Pf CX + X + Pf I + .. ....... I I ( I I I I I I I : Hr Pi Prel .+,0< Co nt rol S ystem N o. 2 I j;'' I I -, I I I I Gf n I a I I i I I I I i I I I I d I G Gir I Gi v ,, I 1 I I I I I r I I I ., --------' '' ,, Indi c ates a co ntr o ller wirh variable dyn am ics determined by a djustments. Ind icat es control component ( sensing elem~nt valve ac t uator or a fixed co ntrol f unction), Indicate a pr oc ea1 performance function, Pr I }. Pn I + p + ~~w p + n ('in Indicate a 1ummin1 point Figure 5 10 Ii pn r;~ r h I 't ..

PAGE 14

CJ) t!) .. t!) f-,.1 CJ) ll.. 11 OHEMICAL EHGil~IIG EDUCATION Dec. 19 62 K Gnv : n and Gnn (1+ T' ns) it is well to proyide in Gn the elements ot K insofar aa practicable. The objective is to provide sufficient gain and to can cel system denominator terms with controller numerator terms thus extending the desired frequency response consistent with realizable components and system sta bility. Physical limitations exist in the control components generally available which implies compromises. Similarly for the injection system the criteria indfcate that Gi Giv Gif should be as high as possible. Once again controller characteristics should tend to compensate for the drooping frequency response of Giv and G 1 r as frequency in creases. In this manner the general requirements of the individual control systems were determined. The next step waa to consider the stability of each loop. Using elementary servo theory the gain in the vicinity of an open loop phase angle of -180 degrees was considered. The gain should not exceed unity at -180 degrees, and some departure from this is desirable. Many rules are available to aid in arriving at the best performance. In this manner a satisfactory value for controller gain was established. The two systems were next tied together by the control functions, Gw and G The first was based on the empirical relation between energyinput and require!P compressor auction pressure. The relationship was of the form 0 -20 -40 -60 0 -0.4 ... Q.6 -KE e------ ""I""' '"fl"!~ 11111rt1 "I -..... 1!1!! l111 'll"tl'I' 1 ---- HIit I I I 1 l ..... l .. r. .... n" ,. 'II" 1r: 111 "" I U 111ru1t liilllt!t!Wi: ,. ... 1 ......... 1n 1 11 .... U I I I :::~ I Empirical performance function determined by curve fitting using profiles corresponding to linear forms: [ s 2 2(.6)s ] .17 (.6)2 + (.6) + 1 2(.8) ] [ s 2 2(.4)s 3 1 5 + l (2.2) 1 + 2.2 -l.O 1.0 O.Ol 0.1 FREQUENCY, RADIANS/SECOND Figure 6 r 10

PAGE 15

Dec. 1962 CHEMICAL ENGIBEEBilfG EDUCATION 12 Also, since the corrective a ction should not occur before the reactor pres sure distrubance appeared at the compressor suction, some attempt was made to create delay time in tbe control loop to match that in the process. A first or der time lag element was used as an approximation. The feed forward element thus became -KE y Finally the block shown as Ga was selected. This control component was pro vided to adjust the compressor sugtion pressure to compensate for slow changes in system characteristics. Gap was thus required to produce an output which was the integral of the input. The form of control f unction chosen for this service was s which was closely approximated by conventional components. The expenimental data were used to enable selection of control components and for arriving at the appropriate ad justments of controllers. For example, a lt h ough Figure 6 shows that the relationship between changes in injection gas flow rate and reactor pressure is rather complex, for part of the synthesis the approximati on can be made that the performance function is first order with a time constant of ab out 4 seconds. (Reciprocal o f the b reakpoint frequency ot about 0 .25 rad /sec) .T hi s means t hat a v a lve actuat~r ha v ing its lar~est time con stant equ al to 0 .4 seconds or less will be quite satisfactory. Thus pneumatic components will suffice. In additio n, the time constant ~ ,. .. in tbe integration component, Gap, shoul d be 1 0 seconds or more if isolation of the terminal control system was to be acbieved in some measure. The integration would then not be effective at frequencies above 0 .1 radian per second. The compressor suction pressure control required high performance components. Moreover, positioning of a lar g e valve to make small changes in a large flow rate was involved. This system called for precision components and a hydraulic system was selected. To help compensate for the valv~ actuator and process some derivative action was introduced in controller, Gn Frequentl-y, however, some hydraulic components inherently provide some le ad action so that none need be deliberately added. E xperimental studies on the control gea r is generally re quired to determine if these properties exist and at what frequency. The above approach to the design of the a ppropriate control system is usual ly adequate and optimum performance can ge nerally be attained by minor adjust~ ments during operation. However, if further verification is warra nted, the entire ensemble may be modeled on the analog computer and the best controller functions determined by a systematic study. The procedure for designing control systems for existing plants can be sum marized as follows. l. 2. 3. 4. 5. 6. Study the plant to ascertain t he fundamental phenomena involved and iti ope~ating principles. Design and conduct an experimental pro g ram to determine important performance fnnctions. Based on the principles involved and bearing the data in mind, for mulate the best control scheme in a qualitative manner. Establish control criteria and determine the more exact nature of the particular contro1 runctions utilizing data from the experimental studies. Construct the complete block diagram, or equivalent, in order to study over-all behavior, possible interactions, stability and per~ formance under extreme conditions. If warrapted, extend the study using an analog model.

PAGE 16

:, 13 l. 2. 3. GHEMIGAL ENOINEENINO EDUCATION Dec. 1962 LITERATURE CITED Control Engineering, Vol. 9, No. 5, pp. 79, May 1962 Draper, c. s., W. McKay and S. Lees, Inst~ument Ept1neer1ng, Vol. 2, Chapter 25, McGraw-Hill Book Company, New ork, Nsw!ork (1953) Hougen, J. o., and R. A. Walsh, "Pulse Testine; Method," _Ch_e_m;.;,._E;;;.!!8~----Pr_o_.gr......_. Vol 57, No 3, pp 6979 l 1961 ) 4. Dreifke, o. E., "Effects of Input Pulse Shape and Width on Accuracy of Dynamic System Analysis from Experimental Pulse Data," Dissertation presented to the faculty of Washington University for partial fulfillment or the D.Sc. Degree, June 1961 .... 4 NOMENCLATURE Performance Functions On relation between compressor suction pressure and energy (energy signal in voltag~ signal out) controller function (voltage signal in voltage signal out) bypass valve actuator (voltage signal in displacement out) process between recycle valve and compre ssor suction (valve displacement in compressor suction pressure out) compressor suction pressure sensor (compressor suction pressure in voltage signal out) process between recycle val ve and reactor pressure (valve displacement in reactor pressure out) process between energy disturbance and compressor suction pressure (energy disturbance in compressor suction pressure out) process between energy disturbance and reactor pressure (energy disturbance in reactor pressure out) controller function (voltage signal in voltage signal out) reactor pressure sensor (reactor pressure in voltage signal out) controller function (voltage signal in voltage signal out) injection valve actuator and voltage to pneumatic signal transducer (voltage signal in valve displacement out) process between injection valve and reactor pressure (valve displacement in reactor pressure out) o 1 n process between injection valve and compressor suction pressure (valve disp lacement in compressor suction pressure out) process between reactor pressure and compressor suction pressure (reactor pressure in compressor suction pressure out) prooeaa between compressor suction pressure and reactor pressure (compressor suction pressure in reactor pressure out) .,

PAGE 17

Introduction -...~ ----.. INSTRUMENTATION IN DESIGN Kenneth A. Otto THE DOW CHEMICAL COMPANY MIDLAND, MI CHI GAN Instrumentation and process control has been a basic ingredient in the growth of most chemical companies. Management has long since recognized that application of advanced instrumentation techniques was essential in insuring strong competitive market positions. The esteemed position instrumentation holds today in most companies has been well earned. Its future holds even greater potential since it represents to management one of the basic keys to meeting comp~tition through production of new and better products, increased plant operating efficiencies and the ability to hold closer (and more exact ing) product quality specifications. Economic considerations -.. ,_ .._..,, -, ,.. .-.--=-, _,_ ..... .. ~ ~ ........ ~ .,.,. ,~ Today in a typical petrochemical company instrumentation and control represents a substantial portion of the company's effort. According to data on instrument sales compiled by the Department of Commerce, the chemical in dustry accounted for 16.6% of all the instrument sales in the country. The petroleum industry adds additionally another 12.5 % In total, the petrochem ical industry accounted for about 30 % of all instrument sales, yet during the same time accounted for slightly les s than 10 % of all new plant investment. The Department of Commerce figures also reveal the ratio of instrument cost to total plant cost ranges from 3 to 15 % for the chemical industry. A lar g e and constantly growing percentage of the money spent by many petrochemical companies goes for instrumentation. For example, The Dow Chemical Company usually spends between $ 2 and $ 6 million annually on instru ments of all types. In addition to Dowrs outside purchases, we spend a sub stantial portion of our research dollar on en g ineering, design, and fabri cation of special purpose instruments in addition to a good share of the test and en g ineering labor~tory work done in the company. DuPont released figures recently that showed an impressive 10 % of their total plant investment as being directly attributable to instrumentation, a value of well over $100 million. Truly, the chemical and petrochemical in dustry is more dependant upon automatic control than perhaps any other in dustry. While these figures are impressive, they perhaps fail to indicate clear ly the true spread of instrumentation costs on an individual project basis. A breakdown that we at Dow have found useful is shown in Fi gu re l and is e. correlation of our experience of instrument cost. In this figure, the cost (as a percentage of total direct project cost) is shown vs the total direct project cost. This data, which separates batch and continuous processes shows a spread of 6 8 % of total cost at an 8 million dollar project level. A spread of 8 11 % is shown at a $ 1 million project level for the continuous process. Batch process instrumentation costs are characteristically less, running on the average 4 % lower than for continuous processes Translating these figures into dollars means, fo~ example, that for a 10 million dollar ~lant the installed instrumentation cost can be expected to run between $400,000 and $ 600,000. Design Estimates ... -Investment figures such as these clearly emphasize the importance of careful econonu.c considerations in the control system proposals. In the engineering design area, the ability to accurately estimate the costs of various equipment configurations and comple~e projects is of primary impor tance. In many cases especially in preliminary proposal stages, the en gineering estimates may make or break the ~omplete project; while in other cases, it materially affects selling price, profitability forecasts, and sim ilar areas. Since instruments and controls represent a significate portion of the total investment in process plants, estimates of such costs must be made with increasing accuracy throughout the pro g ress of the engineering project. Data such as shown in Fi g ure 1 can be quite useful in early stages of project study. An alternate method of showing i nstrurnent costs is as a percentage of purchased process equipment as shown in the next figure. Once a project has been studied in enough detail that the major process equipment has been de termined, this figure becomes a more realistic base for estimating instrument

PAGE 18

. lS CHEMICAL EHGIJIEERIHG EDUCATIOB Dee. 1962 costs. From the figure, for a continuous process we could expect instrument ~ costs to be between 30 to 45 % at a one million dollar project level to a 15 to 20% range at a $10 million project level. Again, batch processing in strumentation costs run less; in this case form 15 to 20~. This seems rea sonable in that instrumentation for batch operation is normally not as com plex as that required for continuous processes. Also, production size batch equipment usually is quite large 1n comparison to continuous type equipment and, therefore, represents a larger portion of the plant costs. The data shown in this trgw,.e, essentially correlating past experience for The Dow Chemical Company, is useful for two reasons; first. it realistically out lines instrument costs and, second, it becomes quite useful in the prepara tion of "quick7" estimates and in quick comparison of alternate process layouts. Once a project has been studied long enough to determine the various control systems required, a preliminary cost estimate based on the installed coat of the necessary instrumentation can be prepared fairly quickly. Installed coats at Dow include such items as labor, painting, wiring, piping, and similar items. Current ly at Dow our installed costs for instru ments run between 160 and 170'/, of purchased instr11ment costs. This factor, of course, varies from year to year and is dependent on local labor conditions and the like. In our detailed cost estimates, we additionally try to include a!l costs of an instrumentation syatem such as piping requirements, power require ments, air requirements, panel and floor space requirements anq the like. For, example, if we install a control valve, we will include in the instru mentation cost, the co st of the conduit run, the by-pass valve and piping and the a1r or electrical requirements. These so called ''extra'' costs orten repre sent a sizable part of total. Man-y companies include such costs in elect1' ical '. .. or pipi ng areas and for this reason their instrumentation costs often appear much lower than ours. By contrast, however, our piping and elect rical estimates may appear much lower than theirs Wb:1-le thse figure~ give an excellent picture or where we have been t~r ..oertainl7 do not indicate where we 9:re going 1n instrument costa .. ,. ..... INSTRUMENTATION COST VS. J TOTAL DIRECT PROJECT COST THE DOW CHEMICAL CO MIDLAND, MICH 12---~--------~ --r .. .. r I I ,. .. -1 10 ' ''7t" ., .. .. ... .. ;: ... ; . : ... :-. . . . . . .. .., ,. '1,"i_. '\ i . . I ., . . . ..... ~ .. .~ . . .. ,,. ... .. .: .. .. .;... : .. .. .. -: 4 8 ~ 6 4, 2 ,, . . . . .. . . . . . . . .. . ' .. : ... . ,,. .. . ~ . ... . .. : ---.. ,.. INSTRUMENTATION, 0 /o OF TOTAL DIRECT COST h 0 0 I I 2 I t .) TOTAL I I I I I ) I l t j CONTINUOUS PROCESS PLANTS i . / I . .. D I RECT . . l . ... .. .. .. .. .. .' .----BATCH PROCESS PLANTS ., ,. --I .. r I I l i I .. . .. ... . . ; .. :.~ ... .. ,\ : -. .... : . : .. .... ,_ ;:.. --~ .. ,,,.,. ; .: :,,.:; :. . "" : "' ..... .,,.::-.. .. ... ... ... ., .. :-.:--:.-. --.-~-s .. : / :: :~ ... .{~ :.-: .... ; ..... ;.: .. .: : .. : :._,, ~ .. : \, ;._ .-: -~~ .. ::-::-. -.;.: '\: .. ~ _-. ...... :::"-~.-.-~-.--:::. .. ..,, 1,,. _, ""'' .-,,-# !v . -. .. ; ... .;.:: \' ::-. :. :;-:. :::-~;~ -:: .. \. .. ..... ... -~ .. .. : .~:._.... ',; ,:, .. :.~; .,.:f_ .:-..t:---.r .: ----.:: -~ ....... .. '. ... .... ; .. ~. ,.. ....... : ..... : ,.,: .. ':,. :-1:.-.. ,:: .. . . .. ,t' ,, '', 't'' .. . ,. ... ,.. .. .' _: .... ~. ; '. : .. .. . .. ..... ...... ., I l I I I I C I -----+-----+---!-----+-~ I I 4 ._) 6 7 8 9 CO S T, MILL IO N :., i. J F DC'L LAR j

PAGE 19

< Dec. 1962 CHEMICAL EllGlNEERING EDUCATION I IN STRUMENTATION COST AS PERCENTAGE OF PURCHASED EQUIPMENT VS TOTAL DIRECT PROJECT COST 16 THE DOW CHEMICAL CO MIDLA N D, MICH 60 ----------'---r---,-~----.----.--------,-----" ---' --,-----~ -----L..,.----L------, INSTRUMENTATION, 0 /o OF TOT AL DIRECT COST ----__,;_ -----I t r I 5 0 I ----~-~ f------t-t--__.., ............ ..... ---. r . I : \ I l -~ -,_.. .. .... -. ---+..- ----------~ 1 40 .J CONTI N UOUS PROCESS PLANTS --+-----+-----~ . . I I 30 20 10 .. . : ; .. ~. ;: . . . .. .... :. : : .. .:. I ,, .. . I .. ____ ., ____ _... L-.. .~ -!----.. t I i __ _,__ ___ -. l .. ~ BATCH PROCESS PLANTS ~ ,. __ ... .. .~. .. ... + ,' ,\ .. + ,...,.. ~. .. ; .' 1 t ,, ' 'I'' ',., ... .. 1 ', \ ,, : .... .. ... ,. .. ,. :, : .. . .. .. .. . ., ,. .. .. ,' \ .. ,, : ... : -.,.' ,: ., .: ,' .... . ,. .. \ .. .... . ' ..... ...... :.' ., '. ~ : .: ,, \' .' ...... .. ... :~ .: ' .. .. ,. ,. . ~ . .. ... ,.,, ; ---'--------+------+------+---. -I .. 0 L---;:__ -~ ___,.: ---...l.-----.:L.------L-:-----,;.._-.;_ ---...i....----'-----------0 I 2 3 4 5 6 7 8 9 .. TOTAL DlRECT COST, MILLIONS OF DOLLARS I In general, we can expect a continued rise in instrumentation coats with more and more of the project dollar being spent in the control area. For example: A) For the past five years or so the purchase price or new instru ments has been increasing at a faster rate than for other equipment. Thia price index rise has been in the neighborhood of 7 10% per year for in strmnents as compared to the Marshall Stevens equipment index rise of 4 7~ per year. While the instrument price rise is today tending to level off, the effects of this difference will be felt for some time in increasing instrumentational costs relative to equipment costs. B) Much of the present instrument cost data, at least for The Dow Chemical Company reflects a minimum of analytical type inst:rumentation in the initial design. Usually the analytical instruments (especially the more sophisticated types) were installed after the plant had been in operation for a time. More and more we find today that many of the analytical instruments are being specified during the initial design of the project. As suoh, you can expect their influence to add somewhere between 2 to 20% to cost figures in Figure 2. C) We are starting to reach the point of diminishing returns in the sole use of more instrumentation. We are starting to see where something a little better than more single element or single loop controllers will be required. Today and more so in the future, we will see not only more in struments pei~ plant but more intercoupling of instruments pel' plant. In essence, we are slowly entering the area of more sophisticated control with both analog and digital control schemes playing an important role. I do not believe there is a one of us that does not feel these schemes will result in increased instrumentation costs. I A s way of example, if one ta kes a $ 20 m illion Styrene plant and adds around a : $ 300 ,000 investment in computer control, the total investment in instrumentation goes from a round $8 00,000 to $ 1,100,000 a sharp jump of 31 % yet with other equipment costs remaining substantially the same. I I I I 1

PAGE 20

r 17 CBEMICAL EIGINEERIBG EDUCATION Dec. 1962 )present Development ,. _. S l Today in our industrial processes, where plant measurements are of vital concern to the economics of operation, engineers are making great strides to ward accuracy, reliability, and ve:r-sat111ty of measurement. we are no longer completely dependent on the old standbys of temperature, flow, level, and pressure .Analytical measurements of composition and product characteristics, rather than its environmental condition, are increasingly finding their way into the initial plant design. We are seeing rapid improvements in measure ment transmission, reliability, sensitivity, and sensor accuracy and stability. Important progress is being made in computation of significate information not directly available from individual primary sensing elements and multiloop performance control, accomplished by computers, 1s increasingly necessary. Since most chemical processing is so dependent on proper measurement and control, let us take a qui~k look at our present design trends. Measure ment and control of temperature, pressure, level and flow are without a doubt still the workhorses of chemical process control. With a gradual swing to electronics, and with refinements both in the elements themselves and in ap plication know-how that fully recognizes their capabilities, steady improve ment in these key measurements continue. Magnetic and turbine flowmeters, capacitance, ultrasonic and radiation type level control devices are but a few examples of important improvements in these areas. In the electronic controls, while there is an improving acceptance in our plants, we are still plagued by the many different signal transmission levels. We are forced somewhat to buy a complete system from one manufacturer rather than picking and choosing from several manufacturers as is connnon practice in pneumatic. Of course, the lack of an inexpensive electronic valve actuator is at resent a severe handicap in wider use of electronics. Increased c nsideration is continually being given to the value of con tinuous analysis equipment during the design stages of new plant construction. Where formerly such equipment was omitted in the initial design due to delays and lack of confidence in reliability, performance, and the like, it is now installed as an integral part of the usual instrumentation. This is not to imply that analyzers are as common as pressure guages for example, but it is routine to inquire into the possibilities presented by their use in design of new plants. Further experience with analyzers and increasing dependance on them is largely responsible for this change along with improved equipment techniques and sample handling techniques. Vapor phase chromatography is the outstanding development of recent years in the analytical measui"ement field, and is well established in many applications. While still used primar ily for measurement today, its demonstrated reliability is leading to in, creased use in closed~loop control. Although spectacular developments in VPC have tended to overshadow those on other fields there has been conside:r-able activity and improvement 1n infra-red, in micro-wave and mass spectroscopy, and in related areas. In creased usage and application of these wide and diverse analytical techniques will add measurably to the instrument engineers design capability. D~amio Considerations I'd like to switch now to another facet of instrumentation and control that I feel is becoming more and more important in our engineering design. For several years now, the instrumentation and control engineers have been hammering away at the importance of process dyne.mies in the understanding and application of control systems. Today a good many engineering groups in the chemical industry have fairly well equipped (but poorly staffed} analog ... simulation facilities. This effort to me symbolizes and characterizes one of the most important advances in the ins trumentation and control effort. For the first time the control engineer has the tools and techniques with which to intelligently compare various control schemes, to evaluate the performance and interactions of such control, and to, at long last, come up with ration al justification and tangible economic benefits of instrumentation schemes. Dozens of proposed control and desi gn configurations can be quickly and easily evaluated; effects of different startup and shutdown procedures can be de termined; operational procedures can be studied; operators and plant personnel tra~ned; emergency procedures can be worked out; all ~ade with the assurance which comes only from the intimate knowledge of both the dynamic e.nd the steady-state operation of the process. As a result of active use of analog simulation techniques, we find the so called control engineers entering into more and more of the actual pro cess design. In many cases, they are no longer satisfied with the process as envisoned by the chemical engineer, but may find that to properly control the process radical changes in design or equipment sizing must be made. (

PAGE 21

' Dee. 1962 CHEMICAL EllGINEERIBG EDUCATION 18 By analyzing the proposed process and its control as a system, rather than as separate entities with dynamic as well as steady-state considerations in cluded the control and instrumentation engineers are able to make signifi cant contributions not only to the control of the process but to the entire process aes1gn. For example, 1n an exothermie reaction step, if the reaction temper ature starts to rise, the resulting ability of the control system to prevent runaway conditions depends on the relative transient speed of the heat gen eration mechanism and the speed with which cooling can be supplied to the system. A partial listing of typical determining steady-state and dynamic parameters ia: Reaction Mechanism Degree of Mixing Instantaneous Conversion Instantaneous Volume Instantaneous Thruput Instantaneous Cooling Fluid Flow Rte of Heat Transfer Heat Transfer Surface controllers; Type and Operation Valve Type and Speed Detector Sensitivity Detector Lags Heat of Reaction In the design of a typical process, seldom, if ever, is the dynamic interaction of all these parameters investigated even though they in fluence greatly the control of the plant, the operating capability of the plant, the cost of the plant and the choice of design. In the past, it has not been necessary to include such information because it was usually easier and perhaps cheaper to overdesign the process to avoid problems. In this example, an oversized cooler, extra surface areas, diluents and other exped ients were satisfactory. Such procedures for insuring controllability, how ever, are becoming less and less attractive to the chemical 1nduetry as our competitive race tightens. The sum total of this effort besides resulting in better designed and better controlled plants has been to bring the process engineers and the instrumentation and control engineers closer together. Today you will find many knowledgible process engineers that can discuss at will such concepts as Bode plots, phase plane plots, and the like; while on the other side of the fence, many more control eng1neera can now at least hold their own in such fields as heat transfer, fluid flow, reaoto:r design and similar areas. Thia cross-fertilization of talents has and will continue to result in immeasur able benefits. Su~ary For the past few minutes we have taken a quick look at only a few of the areas of importance in instrumentation and process control, There are others of vital concern that have not been touched on. The steady but rapi d growth of engineering technology has brought to industry many new methods, concepts, and toola for instrumentation and process control. The further application of this technology to our day-to-daJ engineering design problems will con tinue to bring about the improvements in productivity and oost reduction so prevalent in past applications or instrumentation concepts. Costs or instrumentation, I reel, are rapidly approaching the point where management can no longer sit idly back and consider instrUI11entation a nec essary burden of being in the chemical business. Economic justifications, so long a part of' the chemical engineers way of life, are becoming more and more important to control engineering concepts. With the proper motivation, tools, conce pts, and hardware at hie disposal however, I think most in strument engineers oan rise to the challenge. I J

PAGE 22

DESIGN DATA AND THE ROLE OF THE PILOT PLANT ;;_;;;;~ ~~--~----~ .-. ...,_., a-. ~ .....,_ -~-W.L. Larcamp Engineering Department Union Carbide Chemicals Company south Charleston, West Virginia Engineering designs can be divided into two broad categories: (1) de sign~ which will be ma8s-produced, and (2) designs which will not be mass produced. Examples of the former type of design include airplanes, automo biles, television sets and even the proverbial mouse trap. These designs are intended to be mass-produced to provide many identical items. The ap proach to this type of design is to develop plans and specifications on paper, and then to construct a full scale model of the design prior to any manufactur ing operations. This ~samp1"e'' of 'the design is then subjected to many perfor mance tests to determine if the design target specifications have been met. Based on these performance tests, the design can be modified as required before 1 t is f1nal1 zed and mass-production operations are begun. Al thot2gh the research and development costs for a mass-produced article can be quite high, the suc cess of the design itself is assured in advance of the full capital investment in production facilities, which is an important business factor. Examples of designs which are not intended for mass-production of the de &~ include cha mi cal plants, power plants and even fixed structures, sucfi as oI'"ITce buildings and bridges. This type of design can be called custom design and is characterized by the fact that a full-scale model is not feasibie-the prototype and the final design are one and the same. The design, as an entity, must be committed long before it can be tested. From a business standpoint, this means that the full capital investment of such projects must be made in advance of a pr~ven design. It is somewhat like "buying a pig in a poke". In certain fields of engineering for example, structural engineering this sit uation does not pose any real engineering risks. There is always a degree of personal suspense for any custom designer, but the architect or civil engineer rarely has BDY doubts about the engineering sufficiency of his design. This is true for several reasons: 1. The number of engineering variables is small, and the fundamental relationships among these variables are well established. 2. The properties bf the building materials are known. J. Each custom design is so similar to existing proven designs that past experience and engineering judgment can be fully utilized to compensate for any laclc of theory or precise data. The situation is far differ~nt for the chemical engineer engaged in pro cess design. The number of variables involved in chemical reactions and che 1 ical processing are many, the properties of the materials handled (particularly in mixtures) are not always well known or predictable, and the science of ehem ical engineering is still so young that many of the engineering fundamentals are not clearly understoed. The chemicB.1 proce ss designer thus approaches a design problem with these ge-neral technological limi tation s: l. There 1 s no pract ical opport-nity to build a ''sample" of the design to determine the sufficiency of the design in advance of its commitment. 2. There 1s frequently no assured way to "calculate" the design based on engineering fundamentals, either because the fundamental data are not available or because the engineering relation. ships are not known. In order to minimize these technological risks, the process design engineer must have certain ~hecific experimental data on which to base a pro cess design. To a point, e can over-design to compensate for the unknowns and assure the target design performance. However, liberal over-design may result in a technical success and an economic failure, which in the chemical industry means th!, !!esign__ei: fai+ed. That fact should always be understood and appreciated by the stu~ent, the educator and the practicing engineer. Re search in glassware and bench-scale development studies provide considerable information about a process, but frequently do not provide sufficient data to allow reasonably certain extrapolation to large-scale facilities. The ultimate in experimentation to assure target performs.nee is to imitate the designers of mass-produced article and build a model. However, economics dictate that this model must be a small-scale model because the design will not be mass-produced. The designers of mass-proaucea articles, such as airplanes, also resort to the use of scale models in some areas to save time and reduce development costs J prior to bui~ding an expensive prototype for final tests, --19

PAGE 23

Dec. 1962 CHEMICAL ENGINEERING EDUCATION 20 The small-scale model employed by the chemical industry is called a pilot ~ant. A pilot plant is basically a miniatlll'e chemical processing unit, althou it need not be a complete assembly of the entire process. Some of the steps of the process may be by-passed in a pilot plant design because they are well known and can be scaled up by classical methods. The primary role of the pilot plant is to supply data to permit the process des1gn engineer to span the gap between bench~scale equipment and commercial processing raoilities with reasonable assurance. The term, ~emi -!'orks plant, usually denotes a complete miniature of a large-scale chemical process~ w n.Icfi is used to actually process chemicals from the basic raw materials to the finished product or intermediate. Semi-works plants are nearly always operated as miniature production tJni ts (even though useful design data are collected) and tend to approach a prototype of a larger-scale plant. ADVANTAGES OF PILOT PLANTS .. __ .., --~ Many benefits can be realized by utilizing a pilot plant in the develop ment of a design for large-scale processing facilities. The obvious advantage ~ is that an opportunity ls provided to test a model of the process in advance of finalizing the design, in much the same way that models of mass-produced designs are tested. Of course, the pilot plant is a small-scale model, whereas the prototype of a mass-produced design is a full-scale model. Nevertheless, the pilot plant performance can serve as a useful preview of the commercial plant performance. A pilot plant ideally should be "designed" in essentially the same manner as a lar g e-scale plant is designed. The ava5 lable bench-scale data sl i ould be analyzed and used as a basis for the pilot plant d esi g n. Whenever reaction meche~isms or other performance characteristics are unknown, they should be p o stulated from the bench-scale data and theory to h elp predict the pilot plant performance. This approach forces the early assumption of a mathematical model. After the pilot plant is built and operated, actual performance oan be compared to the design performance. Any departures from the expected design performAnce of a pilot plant should be examined and explained; otherwise, the pilot plant dat a provides little more than empirical, pseudo-geometric scale-up factors, Analysis of pilot plant data can thus prove useful for establishing a true mathematical model of a proc e ss, which can be used to calculate the plant de sign. If the predicted performance of the pilot plant is realized, the original mathematical model assumed for the pilot plant design is confirmed. If not, the mathematical model must be adjusted to reconcile the bench-scale da ta and the pilot plant data {and still be consi s tent with theory). These objectives cannot always be realized in practice; nevertheless, the approach shou ld alwa ys b e atte1npted in order to ob ta.in the maximum potential benefits f rom a pilot plant program. It is often difficult to obta in consistent data on process efficiencies a nd losses from small, bench-scale equiptnent. In many cases, the overall ec onomics of a chemical process are very sensitive to the useful efficiency o! the raw materials, and the financial success of a project might depend on whether the plant efficiency will be, say 90 percent or 95 percent. The pilot plant serves to add a needed measure of precision to the material balances (and also heat balances) estimated for a large-scale process. The pilot plant serves to alert the designer to many potential problems that might not be anticipated or seen in bench-scale laboratory equipment. The operation of a pilot plant with "real equipment" over many hours will un cover such things as fouling of heat exchangers or other equipment, the forma tion of residues, and the effect of minor contaminmtbuild-up in the process. The instrumentation of a pilot plant often closely resembles that of a large scAle plant, and control problems can be recognized in advance of plant oper ation. All of these pilot plant general observations are helpful in estab lishing a successful final plant design. A very important corrollary benefit of the pilot plant is in the area of corrosion. Pilot plants can, and in most cases should, be built of the same materials of construction being planned for the commercial plant. Whenever possible, specimens of other feasible materials of construction should be placed in the pilot plant equipment to determine corrosion resistance under simulated service conditions. Laboratory corrosion tests can frequently be misleading, as any metallurgist can attest. In some cases, it is advisable to construct identical parts of the pilot plant from two different materials, one believed to be a conservative material (and the most expensive) and one believed to be just adequate for the service. The wrong materials of con struction can by equally as disastrous as a poor process scale-up. Conversely, the choice of ultra-conservative materials of construction represents a type of : over-design that can erode the profitability of a prooess. I I I I

PAGE 24

21 CHEMICAL ENGINEERING EDUCATION Dec. 1962 PITFALLS OF PILOT PLANT S ~ f ~_., Cl 41111 ~... .,..,,,._.__ ~ .. "-----~In order for the pilot plant to fulfill its objective of providin g useful design data for the transition between bench-scale facilities e nd lar g e--scale commercial facilities, the pilot plant should be desi g ned at some reasonable intermediate scale. As a bare minimum, the pilot plant should be scaled sn order of magnitude above bench-scale facilities, say a factor of 50-100. The exact factor depends on the plant scale-up and the nature of the process. In any event, the pilot plant is usually large enough that its erection cost alone represents a considerable expense. In addition, the operating costs are high. Most pi~ot plants are operated on a continuous basis, and operatin g labor must be prov -ded around the clock. The design, operation, and analysis of results for a pilot plant involves several engineers and many more operating technicians for a period of one year or more. Thtts, the total costs of "pilot planting" are r 1ui te high and considerable time is expended. These costs {both time and money ) are expected to be recovered by a more economical plant design. How evei,, because pilot plant cos ts e re appreciable, there is a na.ttlral tempt a ti on to tave shortcuts and minimize these development costs. The net result can lead to ~~p~rfi_2i_al pilot plants. Unless a pilot plant operation is carefully planned ana execu ed to secure the maximum potential benefits to the ultimate plant design, it can lead to routine operation and token design data. A pilot pl a nt should never be allowed to become a stereotyped phase of a precess dev elcpment program. S ome p ilot I)lants are built for tr. e development of Er.oducts that are not charncterlzed by specific chemical properti es. For example, synthetic resins 8n d fibers require such empirical properties as melt points, clarity, d yeabil i ty a n d h and''properties tk1at result from processin e techniques and that are not a lways related to chemical composition. These types of materials are not always amenable to production in bench-scale equipment and a pilot plant, or n1iniat,1re plant, is required. Such a pilot plant is utilized primarily to produce materials or establish the necessary processing techniques. When the time comes to design large-scale processing fa cilities, it often comes a s a rude shoc k t ha t this t)pe of "pilot plant" has not generated fundamental de si g n data for scale-up. Careful planning and advance engineering studies are necess ary to avoid this pitfall. 'l r .. ere is a tremendous financial incentive to quickly get a new product to mRrket or a new process onstream. In order to compress timetables, design work on t h e commercial-scale facilities is sometimes started in advance of complet1on of tr1e pilot plant work. If delays in the pilot plant program oc cur (as frequently happen for any experimental work), the plant design may be essentially complete and waiting for the pilot plant data for confirmation. "''hen this !t appens, there is a great tendency to "prove the tentative design" and thus defer or by-pass the remainder of the fundamental design data. Such expedient action, while possibly justified by circumstances, dilutes the effectiveness of the pilot plant and eliminates the opportunity of exploring alternnte process designs that might be more economical. Thus, a pitfall to l: > e recognized in the early planning of pilot plants is the possibility that insufficient time will be available for completing the program, in which case the full potential benefits of the pilot plant program will not be realized. 3orte modern chemical processes are extremely complex and, with the pres ent stF\te of chemicc1l engineering technology, a pilot plant cannot be "de si g ned11 from the usual bench-scale data. In this case, it may also be unlike ly that a large-sea.le plant can be "designed" from the pilot plant data. A rigorous approach to the de velopment of these complex processes might require years of fundamental research studies in advance of pilot plant work. Th e total cost of such a program might well exceed the cost of a commercial-scale processing unit. Thus, there are some processes for which a speculative plant design ls more economical than extensive research and development. JUDGEMENT OF NEED FOR PII ,O T PLANTS -----' .. .._ A minimum amount of specific experimental data must be available for any new process before a design engineer can prepare a realistic plant design. Such information as the effect of pressure, temperature, and contact time on the yield and efficiency of a chemical reaction are usually obtained during the preliminary research and development work in bench-scale equipment. Physical and thermoJynamic properties are either available from the litera ture or can be determined experimentally on a small scale. This package of data is typical of that supplied in most design courses in chemical engin eerin g curricula and does permit a basic process design to be calculated to some dl ~R: r~ 1 rh e chernlca.l industry must choose between two alternate ap pro~c he s to chemic~l proo ri ss desi g n: f \

PAGE 25

. .. Dea. 1962 CHEMICAL EllGllfEERiNG EDUOA~IOB 1. Utilize the data available from benoh-scale studies alone and develop the plant design from chemical engineering fundamentals, judgement and past experience. 2Obtain additional design data from an intermediate-scale pilot plant to better est a blis h the factors affe cting scale-up. 22 The former approach saves t1me and reduces process development costs, but the performance of the plant design is less predictable, The latter approach extends the time and costs of proc e ss development, but the perform ance of the plant design is more predictable. Neither approach, of course, will result in a certain design, There are no hard and fast rules to judge which approach is better, and every new proc e ss should be separately eval uated and analyzed to provide a basis for decision. For some processes, an engineerin e evaluation will show clearly that a pilot plant program is not ~a ndatory for a re asonab ly certain process scale up, If the reaction mechanisms and rates are available or can be readily ob tained in the laboratory (as for many slow homogen eous reactions), it is us ually possible to calculate a process design from chemical engineering funda mentals. If the potential mechanical problems that might be encountered in large equipment are judged to be minor or solvable during a plant startup, the pilot plar1t can be s ~1f. ely eliminated. Even if equipment mechanical prob lems are anticipated, there is no assurance that a pilot plant program will provide solutions. Thus, a alone may rule out the need for a pilot pl~nt. Unfortunately, many new processes today are complex ar1d involve such steps as polymerizations, heterogeneous rea ctions and catRlysis areas in which the present state of technology is limited. It is virt Jally impossible to develop a mathematic 1-1l tnod el from laborat o ry c.lat a, and scale-up to commer cial facilities is difi'icult to predict with assurance. Based on a techno logical judgement, a pilot plant m ay a ppear ma n da tory for a s a tlsf a ctori de sign. However, there are other factors t ha t are pertinent in the ultimate decision w hetl1 er to build a pilot pl9.nt. The estime.ted economics of the over all project and the pollcy of the partic,1lar chemj : ca l company weigh heavily in this de c i si on One quant5 tative eval uation tec~1ique is to e sttm a te the prob able ran ge of the econorriics of the proJ13ct with and wi t hout pilot pla11t effort. SllCh ecot1omic analy sis req,ii res gu od-erigineertn 5 jticfgement, bu t it is usually possible to predict the opt ~~i st ic and pessim i stic project economics to a confidence level of, s ay 90 percent or b etter. These estlmates, to gether with the sales 1)otentlal of' the pr oj ect an d the financial 1~-ts\{S a particular chemical cornpan~ ,r ls .-11 lli.ng to nssl1:ne can be us ed q_s a basis for the decision whether to build a pilot pl~nt

PAGE 26

' INDUSTRIAL DESIGN OPTIMIZATION -----Edward P. Bartkus Engineering Department E.I. du Pont de Nemours and Company Wilmington, Delaware Introduction Chemical companie s, like other industrial concerns, are in business primarily to earn money w1th1n accepted ethical, sociological, and legal restraints. Thus, the acceptance and application of any new equipment or system depends on how it will help earn money how much money. This paper, thel'efore, is concerned primal'ily wl th techniques of design or a plant which will produce a product which will earn more money for chemical companiea. Much or the content is based on major studies of the engineering function in the Du Pont Company, and on eng1n eel'1ng computation, operations research and systems engineering programs in the central engineering department. It is important to note that the output of a design project is not a plant, but a product. The concern of management is not primarily with the appearance of the plant, or how much the equipment costs; but, can the re sulting facilities turn out a saleable pl'oduct which can be sold for a profit. Design~ then, must be the effort to determine the appropriate mater ials and equipment which will produce the given profitable commodity. Design must be the integrating process which selects the pertinent elements from an array of possible alternatives. The elephantine number or options gives the design process an artistic, imaginative slant. For. given the same scope of work or plant objectives, no two project enginee rs working independently are likely to produce an identi cal engineering design, regardless of how formalized and standardized the techniques, procedures and equipment may be. Modern mathematics and computers do make possible a logical process of selection even with a massive list of possibilities. Mathematics can general ize design experience, and computers can accomplish the large volume or cal culations implied by the mathematical structure. New techniques like computer control, process optimization, systems en gineering and operations research serve to make design procedures more rig orous. Terms euch as linear programming, dynamic optimization, nonlinear contra1nts, ridge analysis and dynamic progralllITling enter such a picture. But, these techniques are in their early stages of development, they have a long way to go, and not many lmow how to use the techniques effectively. Actually, much of the effort is linked together by the central idea of trying to do something in the best way there is an optimization problem. To repeat, altho~gh scientific knowledge and approaches are used increasingly throughout the design process, the fact remains that many individual j udgroents must be made. These decisions include sequence of and time to produce an individual design, materials and equipment to be seiected, safety factors to be considered, the amount of spare parts to be provided and the like. Such decision problems have led to this discussion on plant optimiza tion. In this paper, the following are discussed on a high-spot basis: 1. The failure of some plants to make money during the first Year of operation. 2. The definition of optimization and "best''. 3. Some techniques in optimizing the design function and process and project design. Wh_z Pl~ts_Fai.~ A sul'prising number of new plants are not successful. (Fig. 1) study by Chaplin Tyler covering the first yearts performance of one hundred projects ranging from new plants to extensions of existing plants revealed the fol lowing: 1. 2. 3. 4. About 4 out of 10 of the projects showed a loss, or achieved less than 50% of estimated earnings. About half of those projects deficient in earnings could trace their failures to faulty market analysis. One-quarter of the project failures were attributable to process or operating d1ffi9ulties. The remaining quarter were due to general business conditions or delays resulting from start-p difficulties. 23

PAGE 27

.. J Dec. 1962 .. CHEMICAL EROIIED IlfO EDUCA'rIOB PROCESS ~--r-,. --START -UP DELAYS ,,/ D lfFICULT IES / GENERAL S US INS S CONO IT IONS POOR MARKET ANALYSIS 50t Figure 1 I Thus technical deficiencies caused one-quarter to nearly one-half of the financial failures. One can ~nly conclude that a better de&igned facility would likely have minimized this percentage. Idea Inception To Commercialization In order to discuss which techniques of optimization are useful, agree ment on the sequence of steps in a product venture is a prime requisite. There are listed in Figure 2. It is important to notice that continuing economic evaluation or ven ture analysis is essential to maximize the probability of success of the project. What Is Optimization? In industrial design optimization, the objective is a formal, reproducible way of evaluating alternative parameters to produce some "best'' design in terms of a predetermined criterion. What is this criterion? Du Pont has an economic objective of a euitable return on total investment. Others use criteria, such as five-year pay-out period, net cash position after 10 years, etc. But once a venture has reached the stage where a plant is being specified, what kind of plant is wanted? In specifying the achievement of the result "best" for it, each part of an organization considers that his sole responsibility is to make his own group appear as though it has been well-managed, and asks for a plant to suit his needs. (Figure 3). Research and process development ask for the purest product, and a well-instrumented plant that will give lots of data, and a plant that can be used for full-scale experiments a complex plant. Manufacturin g demands a simple design, a 100% dependable low-maintenance minimum instrument, minimum unit cost plant, with flexible and 130-150~ latent capacity, but which will produce on~ package of one grade of product so that inventory is kept low. Sales and distribution want a flexible plant to produce any package size of any grade of product on demand and a high inventory of all product package& and varieties to satisfy customers' desires and minimize loss of sales. Management may want a plant which gives maximum profit but which will el i minate or min l mize the risk, or at leas t minimize the regret. on top or all this, the plant must be absol~telz safe. These criteria of "best" are almost mutually exclusive. Achieving each one of these simultaneously is impossible and impracticable.

PAGE 28

. CBEMIOAL EIGIHEERING EDUCATION Dec. 1962 THE STEPS FROM IDEA TO PROFIT IDEA CONCEPTION PRODUCT SCOUTING RESEARCH PREL IM VENTURE AMAL YSIS PROCESS Rfi. PRODUCT REt MARKET DEV. VENTURE ANAL I I BE"4CH SCALE PROCESS DEV EQUIP DEV. MARKET DEV VENT ANAL. DES DATA DEV I I S / W PILOT I I PROCESS DESIGN VENTURE ANALYSIS PROJECT (DETAIL ) DES VENTURE AHAL YSIS .. I PROCUREMENT COt-lS TRUCTION ST ART-UP TRIAL RUMS ,. PRODUCTION SAlES I MAI( ING MONEY 1 CUAL. IMPR 1 BOTTLE HECK REMOVAL EXPANSION MEW TECH MAKING MORE MONEY Figure ..... .. .. ........ .... .. ..

PAGE 29

, nee. 1Q62 CHEMICAL EKGIIEERING EDUCATION 26 MAX. PROFIT vs. MIN. UNIT COST SOME ''BEST" CRITI:RIA FOR PLANT DESIGN I RESEARCH & DEVELOPMENT PUREST PRODUCT SALES I UN IT COST I MAX IM UM FLEX I B I l I TY MAXIMUM COMPLEXITY FULL SCALE EXPERIMENTS MANUFACTURING SIMPLE DESIGN 100,, OPE RAT ION MIN/MUM MAINTENANCE MINIMUM UNIT COST CAPACITY VARIATION, 50-lD ONE-PACKAGE SIZE ONE GRADE OF PRODUCT LOW INVENTORY SPECTRUM OF PACKA GES SPECTRUM OF GRADE S HIGH INVENTORY MANAGEMENT MAXIM~ PROFIT MINIM~ RISK MINIMUM REGRET I I I I 1_j\J~~R~. __ c 1 o :~ I MAX. I PROF IT I MIN I COST X y PROFIT --PRQDUCTlON ( LBS. )----i~ Figure 3 Figure 4 Let's illustrate with a .simple example. Plant operations often strive for minimum unit cost of product. Engineers design the process and equip ment accordin g ly. The plant, however, should have been designed for maxi mum profit. Figure 4 illustrates the difference between minimum cost and max imum profit. Dollars (profits, costs) are plotted against units of output in pounds. As output is increased from zero, the average unit cost of product is high at low outputs, falls as output is increased, reaches a minimum and then often turns upward. The reason for th is is that at low outputs the fixed charges must be apportioned over a few units of product, resulting in a high unit cost. At medium outputs (between the dotted lines), the same fixed charges are divided by a substant i ally higher output, whereas variable unit costs are still reasonable. As outputs are pushed higher, perhaps beyond originally planned capaci ty, operatin g costs start climbing rapidly. Overtime and night differential go up, higher maintenance is required because operation ls higher than at rated capacity, u nit sales g o up but gross sales dollars may go down because of lower pr i ces from over supply or because of competition. The result is t hat the profit curve may go through a maximux:ri before minimum plant ope ratin g un it cost is achieved. The above exam p le illustrates the necessity of venture strategy planned according to predeterm i ned cr it eria. In a new venture, the best plant or facility is one of a set of alter.' native d esi gn schemes, anyone of which could be developed within the physi cal, economic and social constraints prescribed by management. This plant must be fully described by the set of actual consequences which differs least from the desired consequences, and must answer many of the following ques tions: What ls the plant to do in terms of (Figure 5 ): A. Product (grades, packages, color). B. Performance and reliability (size, duration, efficiency, appearance, ma i ntenance, safety). c C ost and Profits (absolute, relative, or competitive situation.) D. Time {when product is wanted, need for capacity increase.) ( E O ther constraints (risk or regret). Introduction To O ptimization Techniques .. From the point of view of a theoretical mathematician, the optimization of plant desi g n should present no difficulties. However, mathemat i cal and computer approaches in industrial design optimiza tion are still l i mited. The costs of mathematical analysis, systems engineer ing, special tests, computation and simulation, must be included in the c oet of engineerin g Many of the decisions in design, of course, do not as yet lend themselves to mathematical quantification and must be judment decisions.

PAGE 30

27 OHEMIOAL EltGINEERING EDUCATION Dec. 1962 MINIMUM DESIGN C RITERIA PR O DUCT PERFORMANCE AND RELIABILITY COST AND PROFITS TIMI OTHER CONSTRAINTS Figure COSTS OF OPTIMIZATION vs. BENEFITS A B --COSTS __ ...,. Figure 6 As shown in Figure 6, the benefits may accelerate (Curve A) as the de gree of and therefore the cost of optimization increases, or the benefits may taper off gradually (Curve B). (We dontt know yet the answer to this question.) Even if one maintained some degree of optimism, the cost of optimizatior will st i ll be high. To decide whether such sums are worth spending, it is necessary to examine the possible benefits of the proposed installation to decide whether the expenditures are justified. Once this is done, the rest is straight forward. The proposal merely competes with other projects for the available investment money. If the pay-out is good, the more rigorous approach will probably be authorized. If others are better, it may be delayE or not done. One suggestion frequently advanced is a trial installation under a research basis to find out what benefits result. Fortunately, many ideas compete for the research and venture doltar, so the program with the payoff is authorizedFurther, rigorous optimization may cause project time delay and as such may be rejected. It is worthwhile to illustrate at this time the complexity of a chemi cal plant project (Figure 7). Without trying to discuss the chart in any detail, one can see that over-all plant optimization is a formidable problem whether or not mathematics and computers can ease some of the complexities.( In view of such complexity in industrial design optimization, technique~ within three areas dealing with a project appear to justify systematization and discussion. 1. Engineering function, or how the project is managed. 2. Process design, which deals with selection of the equipment to ppoduce the product. 3. Project or detailed design, which specified the mechanical aspects of the equipment. Some of the techniques will be covered briefly, and others will be described in a little more depth, primarily for purposes of emphasis and to illustrate the present state of development. No pr9cedure is explored enough to make possible immediate application. This, too, has its purpose in most instances, the t 1 echniques have a long way to go to become standard. Optimizing~!he Design Function I Optimizing process design and project design carries with it the nee~ to optimize the design or engineering function how best to manage the project. Good management or project engineering divides itself into two parts: {A) Using the proper, economically justifiable tools which get the job done. (B) Allocation of resources to match the cost and timing of the project as established by management. Several techniques in the first group are: 1. New methods of data an4 information storage and retrieval. 2. Scale models. 3. Standards. 4. Analog simulation. And the last requires application of modern planning and scheduling techniques.

PAGE 31

Dec. 1962 iii z Q ca Cl) t~Cl) ..I f : w C) ::, a: !;i z IC) 0.. oo u z w~..J~~ !:: 0 5 ICl) IL CD Cl) a: -N tfj ., MOTOR & TURBINES C) ~u ...... zw f w :l ::, a: t; z J 4 u a: Iu W J IIJ ~~>. ~(<'-1}~ <"~ ~,s,..,.. PUMP 8 COMPRESSOR SELECTION RASE H.F., AND BARR O W, M. H z zi 0 Q_ t IIurn ::, ~IIJ a: ICl) III) ca z 0 u VESSEL DETAILING VESSEL DESIGN PROJECT Et~GltlEER l l'JG OF PRO C ESS PLA NTS J. WILEY & SONS ( 1957) CHEMICAL EBGlNE.JmING EDUCATION The Proiect Engineer ECONOMIC &MARKET ANALYSIS RESEARCH Bench Scale PILOT PLANT PROCESS EVALUATION MANAGEMENT OR CUSTOMER PROJECT ENGINEER ENGINEERING DESIGN a DRAFTING Cl) ::!J n i en b f.J.;)v ~'i; !c, b t,~ ""' ,'< ciJ tJ~ b ~O ~'l:-0 ,._,:fo .... ~b i, f-'1-~v ,.,_..@ 'I,
PAGE 32

29 CHEMICAL EHG~IEE:RING EDUCATION Dec. 1962 PROJECT COST VS. TIM ING CRASH I I I I EXTENDED .... I V\ 0 I (.) "BEsr t; I l.&J I a I I 01:: I 0. I I I I I I I I I TB DURATION OF PROJECT-~~ Figure 8 Tools uaed On The Job l. Information Storag~ and Retrieval Computers are particularly useful for this activity. A com puter library of physical, chemical, and thermodynamic data will pay off fully in manhours and savings. In cataloging of similar drawings by number and description, one can effect substantial draft ing savings where repeat design is done. Such a course of action requires only managerial vigor. r 2. scale Models Use of models to depict the process arrangement, piping, electrica+,instrumentation, and heating and ventilation, has re sulted in many advantages. Such models eliminate the need for preparation of piping arrangement drawings. They permit a more thorough review by plant production and maintenance people as de sign progresses. It is much easier to spot and correct many po tential safety and fire hazards and interferences. Models permit better planning by the construction forces for sequencin g of installation of all components, and, of course, are very helpful in training of plant operators. Models have provided unexpected bonuses in foreign work in that they assist greatly in overcoming the language barrier. 3. Standards Use of Standards is an excellent cost cutter, since they repre sent the best lmown practices. In turn, maintenance costs are re duced since plant maintenance forces help develop the Standards as well as use them for maintenance of the plant. It has been found that economies gained from common agreement on certain standard varieties more than offset costs of "bending" of design to meet availablematerials. 4. Analog Simulation This technique has proven to be a valuable tool in analyzing expected operation of a plant prior to start-up. Although it takes considerable manpower, time, and money to simulate a process, it is often justified for portions of the more complicated continuo u s pro cesses in whioh there are many recycle streams and many variables. Simulation gives a good check on the instruments control system and permits going through start-up, and steady state operation. It is quite easy to introduce upsets of all kinds and to check the ability of the control systems to restore operations to normal. Such trials make possible the discovery of shortcomings which would otherwise not be found until the new plant has been started. _./ Justification for the analog simulation then comes from reduced time for startup of the new plant to obtain the desired capacity.

PAGE 33

Dec 1962 CHEMICAL DGI Io EDtJCA!Io 30 I .... 0 l 2 3 Allocation of Re so ur ces ---------CR IT IC Al PATH METHOD ........ 4 5 6 7 TIME Figure 9 8 9 10 ::c V, z 11 Project mana gem ent has to balance resources against demands Experience has shown that there is a minimum total cost for the correct project time project d esi gn/con str uction relat ion s hlp, as s hoWli in Figure B Total cost goes up if the project timin g is increases if mana g ement calls for an accelerated management also has to bala nce the t1 me and cost equipment optimization against the benefits. extended, and total cost plant start-up. Project require~ fori process and . One new tool increasin g ly being used by desi g n or g anizations is the C ri ltical Path Method (CPM). The PE RT system of the Navy is a similar tech nique. C r i tical path plannin g an d scheduling (Figure 9) is a project man~ agement tool t hat leads to the best combination of timin g cost and equipment o ptimizat1 on in projects. The technique is based on an arrow diagra~ -:! which records the logic of the problems in a g riaphical manner, and states th~ pro je ct activities in the sequence in which they are to be performed~ T~e 13,rrow diagram provides the planner with the followin g benefits: : :. :. --1. It provides a disciplined basis for planning the project, 2. It provides a clear picture of the project scope that can be easily read and understood. 3. It provides a vehicle fori evaluating ultimate strategies and objectives. 4 It pinpoints responsibilities of related departments by showing the interconnections among the jobs. Basical ly, the method lets the manager know precisely which steps in the project arie crltica.l to completing the proj.ects on time. He then "bird d o g s." The noncritical steps have some leeway or "float time." So, if they are late in being finist1ed, the completion time of' the entire project 1 s un affected There is no "float time" on critical steps. Obviously, then, t hese steps are on the critical path. The critical path method is different .f rom othe r metl 1 ods, such as a bar chart, because it separates the planning an d schedulin r V CPM not only permits optimizing manpower among several projects, but presents a road-map which matches rigorous approaches versus time and bene fits. PROCESS DESIGN OPTIMIZATION -,, -----!i_o__~ -~~-~~...R~-~~ -: ~I ~~~~ t~~~ess? For the process industrie~, capital investment costs have increased per haps six-fold over a 30-year period. Such increases have made mandatory care ful look-see at just how much capacity is aetually designed into a plant spec ified for a nominal figure. . ., .. .., . .

PAGE 34

z :::> t:; # 31 CHEMICAL EIGINEERIIG EDUCATION Dec. 1962 MARGINAL ANALYSIS 'Z'' V'l 3R '-.!> :\~ z ll ;"\
PAGE 35

V, 01: < :j 0 0 i 30 20 1 0 Dec. 1962 CHEMICAL EBGlli&:ERtlG EDUCATION 32 highest value of this smallest profit selected. This criterion ma ximi zes the minimum profit or maxi min 2 Ma.ximax Criterion With an optim i stic and speculative strategy, the largest profit for each a ltern ative is examined under any contingency, and the altarr:iatiye selected which has the highest value of the largest profit or maximax 3. Maxim C r i terion. If the strate gy is neither comp letely cons~rvative nor complete ly spec u l a t ive, but somewhere in between, a cr i terion meas u re can be maximized using any proportions of the maximin and maxlmax criteria. For example, one mi gh t feel 7 0% conservative and 30 % spec u lat i ve; or, one might be the reverse, 30 % conserv a tive and 70 % speculative, in which case the v a l ue s wo ,.1 l d be diffe ren t 4 Minimax Regret C r i te rion Under this strate gy, the maximum potential losses are dete rm ined for eac h alternat i ve and cont i n g ency and that alternative i s c ho sen whic l"l has the smallest potential r e g ret. P rocess Op tim i zation Indiv~ual E g~ipment: As previous ly stat ed, cost and time of en g ineerin g the "best" de s ign mt 1st be balanced against the significance of the equipment b e in g in vestigate d in terms of its proportionate part of the inve stment, the chanc es of reducin g that inve stment, a ch ie vin g more reliable operatin g equ ipmen t, an d meeting t h e pro ject schedule. There is a spectrum of d esi gn optimlzation approaches v aryin g from su p erficial, qualitati ve, to t he mo st analytical qu a ntit a tive. Wide S et of A lternatives ------------One c an use charts in sele ction for the be st amon g a set of a lter natives Take a liquid-solid-separation system, for exe.mpl e, a.nd particularly a f i lter. The process de s ign en gi n ee r must cons i der both the perforrnance as ~elate 1 t o materials handling fi ltrate quality, cake qual i ty, etc., a s well ~s econom ics includin g ini t ial cost, in st a llat ion ope rat ing maintenance an d re p lace ment cost. Filter manuf a cturers do supply certain perform a nce inf orm a t ion b ased on de si gn t~stin g and experience. They can ass i s t in deve l opin g a cost picture. Ho wever, the de s ign en gi neer is responsible for the system de si g11 The design e ngi neer must accumulate sufficient information so that k"le can con s id er the ef f ect of as many par am eters as indicated by the sens iti v ity of the materi a ls being processed. Often several types and variations of equ ipm ent can perform essent ia ll y t he sa me duty To optimize in a se mi quanti tat ive sense, chart s shoul d b e d evelo ped in whic h the performance characteristics of each variation can be 1 compared aga ins t a common base 1 ... RETURN ON I NVE STMENT VS. CASH FLOW DECISION RULES FOR INVESTMENT SCREENING Annual Income A I I I t--1--"11-+-----t. I .. 1. MAX IM IN: PESSIMIST IC, CONSERVATIVE HIGHEST VALUE OF SMALLEST PROFIT 8 I I l .,... I I I I I I I I I -, I l l I 2. MAX IMAX: OPTIMISTIC, SPECULATIVE HIGHEST VALUE OF LARGEST PROFIT. 3. MAX IM: X"/o CONSERVAT lVE Y"/o S PECULAT lVE. 4. MIN IMAX REGRET: LEAST REGRET. 0 ------'------L-----'----.L-------J l s 10 1 5 YEARS 20 25 Figure 13 A B C RETU R N ON OR I GINAL I NV ESTMENT l 1e. 12% 12% RET U RN O N AVERAGE INVESTMEN T 2$ 2'1/o 2~ RETURN BY DISCOUNTED CASH FLOW METHOD 2410 15. 5% l~ I:il. J.. 1gt 1 re 12 Figure 13 J

PAGE 36

1000 too 100 700 100 &00 -400 300 250 t3 O 200 I.J.. w a: J.&0 cl :::, g C: 100 w 0.. 90 w a o 70 60 > 50 _j .. 0 w 0::: 30 20 33 CHEMICAL ElfGINEERlNG EDUCATION Dec. 19 62 HOI IZO TAL V GUU ,11 Tt 3' 01 TYi C 30-4 ST. : -4. 6 D A I -6' >IA. SMITH, W. C, a GIESSE, R, C, J1Q8 '.61. v .. H '" ,,; '(( I r ........ R IN D, ENG CHEM,, 53, NO, 7, P 1''40 C ,RBON ST[ I t --i-... (JULY I 9 61) ...... l'i -8' D A VACUU Dlfl ur _T J'x I' ......... TYf'I 30-4 ST. T. -......... Id IA h ---............ i-.. ......... ....... 1 12' DIA ....... i'F ; )( 3' .......... 1" 5'x: v.-.cuu, D~U ... ,,, TE ..... : ..... ,ARB< N S "EE I ............ I '-.. -...._ :'--1 r--5 ( 4' -............ 1 ~I!,' )IA ............_ I ""' I -5' 5 ........... ... I I I ............... 1 ORll,NTA "LATt PR;SS ILT~R I 1 l r-s'x6' I TY l o, 3 1 8 $T T. ..... D SK f" IL.: ER ............... I I I r-SX7' TUB F" L.TER I r i ~AA 80N '1'[ 1-....... j I -...., I xs ............... I 1 ~ : r r 6 8' 10 CARBO STt I I I I I I r 6'x7 .' (rlLTR~ ~ IO N) ICiL. PRESS f"I l N r e'X8' -ax12 J 1 1 1 r :--........... '\... ~! I r 10 x,o I VtR 1 'I I ., I I I _. i .... ............. ...... I I I T Pf! 304 ST 1T, '" ....... ..... r-.... I I l I i----.:... -8')(8 I ,n x r:i' \ ...... I L TER ............... I I I "-... ............... I'--.. .......... I I I ._ I 10 14 T U BE F" ... I I i-... r CARBO STE~L ............ ; ~V I ~ "~ l I I I ::..4. PRESS r1L ........ : 6 ,.... RT I ER .... J / 6X (C L.A RI ICATIC NJ '-...._ ....._ ...._ C ,R ION STE "' "-.... I -.............. .......... I I I'-.. ........... I I I -....J. l ..... ..... l'-k I ...., ' ...... I ...... "' :-............. "'---.._ I I I --..~ I !'-,,,.,.. ............... I ............... ........ l "'...__ I I r---.. ....... ....... .......... r-,.__ ,' I .............., -......... ....... ......_ l ......... I I'-......... I I I ............... '' I .......... O"IZO TAL It~ S I ILT '" ............. -.............. .... r..........::: ......... T Pt J .. sir. s ........... r----..... .... ........... ..... I l l ............... ...... ......... -~ ....... -I ............. ....... NOTE : r---......... ..._ .......... I ALL VALUES BASED ON FILTER AREA ONLY-NO ACCESSORIES ....... .......... i""-........ .. ....... ............... ONZO lAl. t"E: s ... ~" ......... S1 :ti -:,.... I'-.... C "80 2 IN EACH CASE VALUES ARE BASED ON THE MOST FREQUENTLY ----....... .......__ ...... USED DESIGN MODIFICATION ............... I'---. I S I 7 I I 10 IS 20 I& lO 40 so 10 10 10 to 100 I &O 100 210 100 400 100 a00 7'001C)0900 IOC COPYRIGHT 11101 B'I' AMERICAN CHEMIC"L SOCIETY AND REPRINTED BY PERMISS I ON O F COPYRIGHT OWNER, TOTAL FILTER AREA-SQUARE FEET Estimating Cost Chart Fig ur e 14 --1 THE TEXTILE FIBER FAMIL y TREE l Animal Fur Hair Silk Wool NATURAL FIBERS Mineral Asbes1os Vegetable Cotton Hemp li nen Sisal S. EHLERS, I ND ENG, CHEM,, 53 N O, 7 P, !!!!2 ( JULY 1 0!!1) COPYRIGHT 1 96 1 BY AMER CHEM. SOC, ANO REPRINTED BY PERM IS S I ON OF COPYRIGHT OWNER, Synthetic Acetote Acrylic Madacrylic Nylon Ny1ril Oleftn Pofyes'ter Rubber Soron Spandex Vina l Vinyon ) /v' AN-MADE FIBERS f Regenerated -----.-------' r Cellu l os e Gums Rayon Rubber Mineral Gloss Metallic Pro te i n Azlon 0 G,u,eric names lor man-mod e Jlber by Jh Te,cfi/ e Fiber Prod uc t, ld enti flcotlon Act opproved September 2, 19 58, mod effective by Federal Trade Commission Marc:h 3 1960. I F'igure 1 5 I (

PAGE 37

. nee. 1962 CHEMICAL ENGilfEERIBO EDUCA'rIOW 34 VARIABLES IN FILTER FILTER CLOTH SPEC IF ICAT IONS FABRIC CONSTRUCTION GENERIC TEXTILE FIBER CROSS-SECTION CRIMP YAM-TYPE MONOF I LAMENT MULTtf tlAMENT SPUN-STAPLE YARN SIZE AND WEIGHl TW 1ST AND PLIES THREAD COUNT WEAVE PLIED SINGLE PLAIN, TWILL, SATIN FABRIC FINISH Figl 1re 16 MAINTENANCE MAXIMUM PRODUCTION RATE MAXIMUM SERVICE llFE MAXIMUM GASKET ING MAXIMUM SOLIDS RECOVERY MINIMUM BLINDING MAXIMUM CAKE RES I STANCE E C ONOMICS PRODUCTION Fi gure 17 PROCESS PRODUCT Van Note, R. H. I & E Chem., 53; 546 11961) Figure 18 This permits consideration of a sufficient number of alternatives to result in a good or 11 best 11 design. An example is Figure 14, an estimating cost chart, in which the relative value per square foot of basic filter types in available sizes and materials of construction is plotted against the total fil ter area in square feet. (3) Detailed Qualitative A_n~1lys_i~ Very often, qualitative investigation in depth is critical to selection of the best material and equipment, a filter medium, for example. The filter medium, organic or lnorganic, maJ be: a. Granules or powder. b. Porous, perforated or sintered sheets. c. Fibrous wove n fabrics, nonwoven felted materials, mats. If fibrous, the medium may be selected from one of a lar g e number of generic types (Fig ure 15) whose ge neral classification include, animal, min eral, vegetalbe and man-ma~e fibers. (4) The variables in filter fabric construction (Figure 16) must be eval uated. (.5) Further, these para.meters have to be cons3 dered in the light of the such desirable fllter!ng cr1aracteristics a s in Figure 17. Some chara.cteristics ma-y suffer jn order to enhance others. And to add to the confusion, one must c onsider the me.ntlfact.lr er of the original fiber and the manufacturer of the filter. Analysis in depth may mes n the difference between success and failure of the flltration system. \.Jhere t1me Al l o \-1s, lab tests, of course, will be of help in screen ing

PAGE 38

35 CHEMICAL EHGIHEE8ING EDUCATION Dec. 1962 .. OPTIMIZ IN G SELECTIO~J OF DRUM FILTER f r--_ ........ l _____ PRE PARE INP!JT DATA l l) l I SC AN INPlJT D .t\TA 12) I S ELECT PREllM CAKE TH l C KNES S (3) I I SELECT p I I (4) ,__ _____ ....___ ___ _, COMPUTE F!LT!~AT I O N T I rv\E I 5 1 I I ADJUST AP, Li ,9f T O ACCEPT ABLE COMB l ~ J A f ION l 6 } I i i ( -t I i I I I I I I ) I I I I I I I CONS IDER F IL TR ATE VAPOR PRESSURE (7) I CA L CU LATE FILT WASH & DRY JNG TIMES 18) ADJUST DRUM SPACE ALLO TM ENT 19) 1 I ALLOT SPACE FOR CAKE REMOVAL f 10) I I CALCULATE DRUM AREA l 11) CALCUl ATE DRUM Dlf\.,ENSlONS l l2 l r --1 I 1 1 1 I CALCUL/ \ 1t DRUM SPEED ( 13) I f I CONS I DER PARALLEL FILTERS l l4 ) l I I ,-_____ ..__ ______ l SIZE AUXILIARIES I I I I 1 I l J 115) -------.--------i I CALCULATE INV ESTM ENT COSTS 1 16 ) -----,---------' I CAtCLJLATf INCREMENTAL OPERATING C OSTS 117 ) F IND MINIMUM V ALUE OF C) P. C OND. &MJN IN C OP. COST i 18 ) -------..-------a .... . -~ -. --~ PR I NT FINAL SPECS 19) Figure 19 Ass1gn1ng comparative Quan titat i ve Valt1e~ .;.______; ::::.-.. --' --A more ri g orous appro,:i.c h could be tb.e use of the Pyramid Mer j t S ystem 11 The procedure is a s follows: 1. 2 3 4 5 S elect tl 1e four cr1 tlcal parameters (Fi g ure 18.) a Process and Product Sto i ch1ometry,conditions yield, product quality b. Maintenance Re p lacement, prevention problems, and cost. c. Production P ro cess ope r ation d. Economics Ef f ect of improved product quality, deprecia tion, interest, taxes return. Prepare a diagram represent.1 ng the four variables. ( 6) E st ablish a mer i t s y stem for each of the dimensions of con si de r ation which en t er into the selection of the equipment Reduce this mer l t rat ing system to nume r ica l values, even if only comparatj ve. Plot the numer i cal values so t hat they can b e ba l ance d o n e aga inst the othe r and an optim\1m s olution obtaine d The perfect solution to an equipment ~ nit prob lem point equidistant f rom the fot1r apexes of the diagram selection is made of the spe cific equipmen t which only would be d efined b y a Ho wev er, materi a lly, ap proeches this point. In this ty pe of evaluation, care must be taken to give e a cl 1 Jte m its p r oper we ight in :relation to i ts ove r-all contribution. Compute_!' O ~ imi za ~ ion of E~lE~ n.!'. In a st il l more r i g orous e pproach, where poss i ble, p a r s 111eter p hy si~ a l property and cost correlations are pre p ared in u se of a computer to d o t h e many calculations required to optimize the r Jesi g n S uc}1 H r. e. p p::ro e cl1 h as been found to be of use ln se l ectjon of a drum fi lt er ( Fi g; ure 1 9 ) Afte r the inpu t data { 1) ha ve been prepared the. comp1.1ter ( 2) s ca ns t t 1 e d~ta and instructions provided, to estab lis h th~t the problem 1s not in determinate for lack of adequat e information ( 3) l'r) les s in stru cted diff erent ly, it next selects a re aso na b le value for cake thickness and a pre ssur e drop (4 ) a.c ross the cake of 1 0 l b ./sq. in and ( 5 ) compute s filtration t ime If the time is outside prescribed bounds, t he compt1ter ,..,ill adjust press\ire drop until an acceptable combination ( f, ) of pre ss\ .tre drop cake tr_ickness and time is found. (7) During t h~s phase of the ca lcul atlon t he effect of filtra te vapor pressure is taRen i nt o account. If, for any rea s on, a su1teble com bination of variables ca nn o t be found, the difficulty i s ldentifieu by a

PAGE 39

Dec. 1962 CHEMICAL ENGINEERING EDUCATIOR suitable typewriter comment and the computer stops prior to acceptance of new data. 36 (8) From the selected variables, the ~omputer next calculates the time required for the filtration, washing e.nd drying operations. {9) The angular allotments for each phase are computed and checked for practicability, with suitable adjustments being made, until a workable assignment of drum space is achieved. (10) During these calculations, appropriate angular allotments are made for cake removal, waste space and resubmergence, based upon accepted practice for vacuum drum filters of various sizes. (11) The drum area is next calculated and the drum dimensions (12) selected from tables of standard filter sizes. (13) Drum speed is computed and compared with accepted practice. If too high, the cake thickness is increased and the entire procedure repeated. (14) If more area is required than can be provided in a eingle filter, the smallest number of parallel filte~s of equal sta!?:dard design will be spec ified. If the selection of a standard designresults in the provision of more than the needed surface, the computer will take advantage of this circum stance to reduce vacuum pumping costs by reducin g the proposed pressure drop to the extent permitted by the surplus area. (15) ~Text, the vacu1 .1. m pump and piping is sized and the investment(lb) required for the entire installation computed. (17) The operating costs for the entire filtration operat 1 on are com puted, including the assignment of an appropriate value to the solvent loss through the vacuum system. It should be noted that this "Incremental mill cost" includes only those items of cost which are functions of the operating and design variables disc11ased above. A full mill cost is not computed nor sho1.,ld it be inferred from the reported results. (18) Unless instructed otherwise, the computer will now proceed to adjust operating conditions until it has found a combination which results in a minimum value for the incremental mill cost. This search is initiated by re ducing the pressure drop until the next larger standard filter will be required. Reduction of pressure drop continues until all the area of this larger filter is fully utilized, whereupon the incremental mill cost for this installation is computed and compared with that for the earlier filter. This process is continued until three filters of increasing size are observed to result in no net mill cost improvement. (19) At this point the computer will type out the cost of the best combination of variables located and stop. VPlues for the operating and design c onditions for all designs explored are displayed, to gether with the applicable investment and mill cost estimates, for ready comparison between cases. Process Flow Sheet Optim i zation Suboptimization of individual equipment items certainly makes for bet ter design and operation of that functional unit or en g ineering operation. But what effect does this suboptimization have on the entire process? If the sequential steps have minimum interaction, other than product inp u t an d out put (for example, with no recirculating streams), such suboptim i zation con tributes fully to obtaining the "best" over-all process. ACETYLENE PROCESS ALTERNATIVES NAPHTHA ~OVIN G AR C: ARC REA C T O R W UL FF F U RN -PR O PAt
PAGE 40

37 CHEMICAL ENGINEERING EDUCATION D ec, 1962 (4.3) (44) (lp) ( 46) (47) (48) (49) P ex l p ex 0 i rl l I rl 0 C rl 0 PROCESS MODEL (REACTOR OU'l'PUT) 2494.3 771 yr 2 REACTOR B l?(l~M, a [1 rl + I orl ] 227~98 Y4F l ~2: ~s r I rl + I rl l 2_2, __ .. ~1&41 C Paa Y r .3 i.Tr 2?, lJ [1 rl +Iorl 227.98 l Figure 21 Ctr 0 22.13 120.14 Should t he operating conditions of some steps b ear on the s uccessful ope ra tion of others, if t her e are recircu lat ing stree n1s if there is inter ac t!on, then individual equipment optimization may adversely affect othe~ pha ses of the process Qu a l itative Optimization To try to apply ri go rous, forma l i zed techniques in all cases would be ridiculou s. For example, consider a case (Fj .g ure 20) w here some hydroc a r bon s uch as acetylene, propylene, or ethylene, i s to be produced T h e probl em is to sel ec t the best process frorn the feas ib le a lternatives, s ub j~ct to certain constraints, suc h as time competj tior :.. plant loc ation, utjJ.ities etc. Assume three kinds of feed stock Ea c h fee d may be pyru l yzed i n, say, eleven different types of reactors. The crua.e stream frorn t\1.e re Ft ctors c & n be compressed i.n any of pertiaJ)S three differe.r..t types of eq'l) ipment a n d p1.1r ified in an y of perhaps five purification trains. Without consi d er in g all external environmental factors tl 1.e 1 ~ e cou l d be 1 ~9S possible p rooess com b1 nations. In today s state of the art, r igo ro,1s approach to selection of the best of the 495 i s out of the question Many of the com b inations are likely to be incompatible and can be so determined on a judgment basi s by exper i e nced engineers. Ho wever, once t he reasonably possible rot1tes are selected, increasingly soph.isticated t echni ques can be used to select t he best process EVALUATION OF ALTERNATIVE~ ( CASE APPROACH) COST MATl4EMATICA1. MODEl. PERCENT REACTION PURIFICATION NO. FEED RETURN INVESTMENT l Fl Partla\ Oxidation Absorption 19 DISTIU.ING COLUMN 2 Fz P1rtl1I Oxidation Distillation 10 PACK ING COST:: lf 6 (H) D2 3 F3 Partial Oxldatlon Molecular Steve 15 1 .. LAIOI COST t ,220, f H>o. ,,,o. im n Fn Pyrolysls Dlstlllatlon 18 3 Figure 22 Figure 23 PERC ACCUR t3 t !5 10

PAGE 41

Dec. 1962 CBEMLCAL EIGIMEERIBG EDtJCA~IOJ 38 Cas e Approach wl th Compute!~ The case method is an optimization approac h. A s experimental work pro ceeds, an d the flow sheets are being developed, the flow sheets are described mathematically, Both hea t and material balances a nd alte rnative process steps a re considered. Here is an example (Figure 21) of a reactor which has eight feeds and one discharge. The set of eq uations is presented a s an il lustra tio n of the type of express5.ons required to specify a process mathematically. Enginee rs do this analys~s anyway, but this approach fo rmalizes the effort. The resultin g equations are put on a digita l computer and these equations varied a s to ope ratin g con d!bi ons Alternative proces s ing steps can a lso b e considered. At the present rate of development of the lo gic, JOO to 500 es sentially algeb raic equations d escri b e a flow sheet. In o rder to be solva ble on a computer, they Must be appropriately ordere d or a rr anged so th a t a solution can be achie ved in a reason a ble amo1mt of computer time. If just ified, optimization mathematics will do the selection quickly and automat ically. ( 7) It is now necessary to introd11ce ( Figure 22) investinent and operating costs so that a l ogica l choice can be made from among alternative processes~ To inject cost and investment information, effective conceptual design is essential. P. ccur'ate fabricatj on cost, desi.gi1, and operatin g information is requ ired The chart typifies two of the equat ion s required for a typical process step. In a manner analogous to thatfor the process mathematical models, expressions representing the cost for each step can be derived. For each step, invest::ne11t or o t h er profit crlteri on ca,:1. be ex.pressed in terms of t,he same variables of time, temperature an i concentration. Next must be found the com patible v al1.1es of the proce ss operating co11d itions and p lR.nt costs which will maximize the retttrn on investment. Once the flo w sheet models and the cost models have been derived for various alternatives, t hese mo d els (Fig u re 23) ca n b e manipulated via com puter to produce an output from ~h ich can be selected the choice processes f'o r further investigation For exa111ple, from this chart, within certain criter ia for such sel ec ti on management mi g ht choose C9se 1 and c~se 2 for further investigation becaus e of the medlurn high ret1.1rns combined with min im11.rn uncertainty about th OSI? ret1,1rns. or it may be that a .qigher degree of precision is neede c i to exercise a lo gic al choice among alternatives It will be necessary then to iterate or repeat the pro Br am until laboratory and cost information have been refined to the desired degree of accuracy. Based on these factors and many others such as timin g risk, and other economic considerations, the appropriate lev el of management chooses one or more courses for further development and eventually for the commercial facilities. ft I I I I I l
PAGE 42

39 CHEMICAL DGIIEERING EDUCATION Dec. 19 62 Th er e are a n umber of more comp l ex optimization approaches in de velop men t by a n umbe r of compani es i nc l uding Cheops (or Chemical Engineering O pt i m i za tio n Sy st em) by She ll De v elopment Company The very fa ct one sees so l i t tle detai l in l itePa t ure shou l d be an incentive for the universities t o pion eer r igo r ous s y ste m s. Pr~1 ect ( D et ~~ ~ d ~~ign) Op t ~~~~i?.~
PAGE 43

Dec. 1962 LAB PROGRAM FACILITIES COl'JECTURE CHEMICAL EIGIIEERIBG EDUCATION I OPTIMIZATION PATTERN ~ / f .._ :s.. -~ s: ...... c :rcrr :cz& a_::e .. PROCESS MATH.MODEL PROCESS COBJECTURE t Comp~1?_e_r Des_is;r; __ ~f E9.,1E-_E,~~nt 40 Heat exchanger costs are often calculated from a family of curves giving cost as a function of heat exchanger area for various materials. costs were adjusted for different pressures, tube lengths, bundle types, number of passes, etc., by means of empirical correlation of actual prices paid. (Figure 2.5) The two examples merely suggest the availability of many other programs for facilitatin g design computation. Such programs are prepared by any eng ineering organ i zation close to a computer. One can obtain many of the pro g rams through the AIChE computer program excha~ge effort or through program sharin g or g anizations. SUMMARY Industrial desi g n optimization can be defined as the orderly consider ation of the pertinent factors which control the possibility of a venture in the selection of specific equipment and operating conditions to produce the maximum profitability. Note the word pertinent not all, or all inclu sive. In application, techniques must be simplified to the point where they will make a major contribution in helping guide research in selection of the proper process arran g ement, while, at the same time, not unduly complicate the procedure at least not beyond the point where the added sophistication ls more than paid for in g ain 1 f carried to the extreme, quantifying tecl1niques would result in exact s p ec if ic at i ons 1,or each component i n the system, to insure some optimum ~ech nical performance of the over-all system. Each component would then be de signed and built specifically to these specificat i ons. However, the engineer ing part of optim.i. zation implies consideration of economic s as well as tech nology so that complete custom designing prices itself out of the market. The final level is an over-all optimization procedure which carefully analyzes all factors and locates all maximum and minimum values of coat and profit in the range being considered. Finally, this leads to a plea for the systems engineering approach. In the chemical process which is operated to produce a product for customers, the static conditions rarely actually are achieved in plant operation. The plant conditions are continually being adjusted to meet the quality and capacity of the demands. Start-up and shut-down operations may establish many of the de sign criteria, rather than the propsed maximum operating capacity. Therefore, since the plant operating conditions are continually changing, the plant de. si g n must be based on dynamic considerations in order to maximize profits. It is becoming too expensive to build a new plant and to experiment with it. Facts must be lmown before process is built not after.

PAGE 44

41 CHEMICAL EIGIIEERING EDUCATION Deo. 1962 This means that the optimization pattern (Fig. 26), leading to an op timum or "best" plant, must as early as possible and as research proceeds, tie in with the laboratory program, conjectures and mathematical models of the process, the prbposea facilities, their controllability and reliability. Engineers have begun to realize tha t they can no longer think of a pro cess plant as a collection of individually designed operations and processes. It is increasingly apparent that each component of a plant influences all others in obvious and subtle ways. Some of the subtle influences are readily found, others can be pinned down only after long, steady experimentation. This is a challenge to professors of chemical engineering, whose contributions can, for example, be as follows: 1. Rase arch to define explicitly the dynamics of chemical engineering operations. 2. Systems and procedures to facilitate the design function. 3. Publications to achieve understanding by industry and especially of elementary, practical, non-complex appli cations of the new techniques, 4. Indoctrination into the college students the fact that graduates who have been a number of years require "time out for mental digestion" of the new, more sophisticated approaches. In this short review of several industrial design optimization techniques emphasis has been placed on getting an understanding on what is optimum or "best," and that cost and degree of optimization must be compared with the benefits. Further, three types of optimization must be considered for a chem ical plant: the process design (or specification of which kind of equipment is to do what), the project or detailed design (the mechanical features of the equipment), and the engineering function {how to get the best job done in the best way). Even though there is increasing use of mathematics and computers, the individual judgments, which can not yet be relegated to rigorous analysis, tell us categorically that design of a chemical plant remains to a large part the Design Project Engineer's or "artist's" personal effort. The challenge is for the university research to maximize the mechanization of those steps which lend themselves to mechanization so that the project en gineer can op timize his time in creating the "best" plant. 1. Rase, H.F., and Barrow, J. Wiley & Sons (1957). right owner. LITERATURE CITED M.H. Project_ ~ng~:r:ieerkn_g ~f_ P_!'oce_s_!! .1 Figure reproauce w!'t permlssion or copy2. Quigley, H.A., and Weaver, J.B., Ind. Eng. Chem., 53, No. 9, 55A (Sept. 1961). 3. Smith, w.c~, and Giesse, R.C., Ind. Eng. Chem., 53, No. 7, p. 540 (July 1961). Figure copyrightea by pe'rmiss!on orthe copyr.ight owner. 4. s Ehlers, Ind. ErJ8...!__C~em., 53, No. 7, p. 552 (July 1961). Figure copyrighted by permisilon o1 the co pyright owner. 5. S. Ehlers, Ind. En6 Ch~m.,.53, No. 7, p. 554 July 1961. 6. Van Note, R.H., and Weems, F.T., Ind. Eng. Chem., 53, No. 7, 546 (July 1961). Figure copyrighted by tlie Amer~ Chem. soc., and re produced by permission of the copyright owner. 7. Bartkus, E.P., E.I, du Pont de Nemours & Co., unpublished paper, "The Systems Engineering Approach to Profit in the Chemical Industry". (2/8/60), presented to South Jersey, Pittsburgh, South Texas, Col umbus A.I.Ch.E. chapters.

PAGE 45

PROGRAMMED LEARNING IN CHEMI C.4.L ENGINEERING EDUCATION L. Bryce Andersen University of Nebraska Lincoln, Nebraska The application of programmed learning to the teaching of high school and beginning college courses has grown at an astounding rate in the last few years. In Septem ber 1962, 122 programs ranging from music to mathematics could be obtained on a routine basis from the nation's publishers (A 3). In addition, several hundred more prog~ams are being prepared. Research in pro grammed learnin g is actively pursued in leading colleges throughout the coun try (Al, A l 2). In view of this intense development of programmed learning in high schools and colleges, the Education Projects Committee initiated a brief study of programmed le~rning as it might apply to teaching chemical engineering. This re port is a preliminary survey of the field, witq a few suggestions for appli cations. Because no programs are currently availa ble in college-level chem ical engineering, the report is necessarily rather general and speculative As the name implies, programmed learning attempts to more thoroughly structure written instructional material. The mechanical teaching machine is perhaps the best known technique of programmed learning; however, the programmed textbook_ may offer mor~ promise for advanced college courses. A third type of programmed instruction utilizing closed-circuit television and a digital computer is in an early stage of development. One autho r ity (A13) believes that the simple programmed textbook (because of its low cost) and the very complex computer-controlled teaching machine (because of its tre mendous versatility and capacity) are the two techniques which offer the greatest promise. Basic Procedures in P rogramm~d Learning I All programmed learn+ng teehniques are designed to present the material to be learned in a sequence of short steps, Usually each step consists of one or two sentences and perhaps a figure or diagram. The learner is most often required to fill in a blank with a word or phrase or to answer a ques tion. Most programs require that the answer be written out either on the program or on a separate sheet. In some cases the answer is fed to the machine by a keyboard or typewriter. Each step requires the learner to make only a small increment in learning and clues to the correct answer are given. As a result, errors are seldom made. A fter answering an item, the learner proceeds to the next step, which wasntt visible when he answered the previous item He re the answer to the earlier item is given, and another item with a question is presented The le ar ner proceeds step..;by-step, answering questions or filling in blanks at each step, until he reaches the end of the program. Programs range in len gth from several hundred to a few thou sand i terns. For example, a programmed textbook on introductory statistics contains 17 00 items and requires 15 to 25 hours to complete. The bas ic sequential pattern of programmed le arning described here has many variations, several of which will be described here in a later section. The proponents of programmed learnin g claim that it has three major advantages over conventional teaching methods (A-6): 1. Programmed learnin g requires continuous, active student participa tion. His response to each question gives him practice at each item, so that each step in the learning sequence is properly learned. It is difficult to be passive or indiffe rent when reading and re sponding to a program. The continuous demand for answers maintains student interest. 2. The student learns whether his answer is right or wrong with mini mum delay. Thi a tends to make him remember correct responses a nd quickly forget erroneous answers. In a conventional classroom sit uation a student often waits several days to learn whether his homework or examination answe rs are correct. Of course, he may ob tain immediate reinforcement of his correct answers from the teacher in classroom discussion, but not all students can participate. Pro grammed learning gives immediate reinforcement to every student just as though each student had an individual human tutor. Educational psychology has shown that immediate reinforcement facilitates learning. A student will remember an answer which he knows is correct better than one he is uncertain of. Conversely, r1 is learning immediate l y th at an an swer is wrong encourages him to forget the wrong response before he has a chance to learn it. Sup plying the correction for the wrong response also aids in learning. J

PAGE 46

t 43 CHEMICAL EBGIBEERIHG EDUCATIOB Deo. 1962 3. Each student ca~ proceed at his own individual rate. Fast le arne rs are not held back by t he slow student, as sometimes occ urs in th e conventional classroom. Conversely, the slower student is a ll ow ed all of the time he needs. Do these claimed advantages actually le ad to better learning using pro grammed instruction? Much research is in progress to compare programmed and conventional instructio n. The results seem to show that programmed learning is a t least as effective as conventional classroom textbook instruction in some areas ( A -13 ) That is, t h e stud~nt learns as much, and in some cases more. However, these conclusions cannot be app l ied to all levels of college courses, bee a use no studies have been made on ad va nced college courses S ince programmed lee rning requires a sequential presentation of infor mation, the material to be taught must be of a type that can be broken jnto a sequence of steps. The most enthusiastic proponents of programmed learning claim that anything tha t can be te.ught can be programmed. t4ore realistic ob servers emphasize the need for defining the objectives of the proposed pro g ram before an attempt is made to write it (A-9). It is necessary to state very carefully exactly what the program is to teach. This is done by stipu lating what the student should be able to do at the end of the program. For example, statement 1 is a much more appropr!ate goal for part of e. higher a lgebra program than statement 2 1, rrhe student must be able to solve .5 p a irs of simultaneous a l ge braic equations in 1 5 minutes. 2 The student shoul d have developed an understanding of simultaneous algebraic equations and the ir solution. How can ''understan ding be measured? Only by requiring an overt respo nse such as that suggested in statement 1. The precise delj neation of what is to be le arned is an i11dispensible first step in writing a program. This delineation must state the overt be havior expected of the student a t the end of the program. If this cannot be done, t here is no point in writing a program because there will be no way to determine whether the student has lea rned any t hing In high school courses, whic h usually emphasize lea rning of sp ecific facts and techniques, the goals often can be precisely defined in terms of overt behavio~! 9n the o~her hand, such definition is much more difficult in advanced college courses It may even be undesire.ble or impossible in courses where the student is en coura ged to tl 1 inl< for himse l f' a n d set hi s own g oals. As a result, it may not be possible to pro g ram many ad v anced college courses. Published programs include only subjects which are well-structured, such as e l ementary mathema tics, or subjects which require the learning of many facts and rules, such as English g rammar, usage, and spelling. Many advanced college courses do not seem to fit into these categories. Mo st of the effort in programmed learnin g to date has been concentr8ted on high schbol courses ( 7th through 12th grade). Much more work is neederl before the value of pro g rammed learnin g in college courses is est ab lished. However, the dramatic success of some high school programs appears to make college studies desirable. Types of ~rograrnmed Instruction Various devices a re betng developed to present the sequence of item s required in programmed le a rning. Three major types Rre teaching ma~hines, progr a mmed textbooks, and computer-controlled devices. B efore these are dis cussed, a few comments on the program itself a re necesse ry. ?rograms and Programming: The program is the heart of pro g rammed instruction. W r i tin g a pro gram is a difficult chore. A lthough several hundred programs have been written, there seems to be littl~ consensus on how it should be done. A programmed primer on the subject gives a few su gges tions (A-10). It is also an inter esting example of a programmed textbook. Most psychologists feel th at learn ing is more efficiently ac compl i.shed by reinforcing correct answers than by correctin g erroneous answers. A s a result, most programs are written to elicit a correct response. This is ac complished by providing sufficient "cues" to the st udent. The answers to the first several items on a specific subject may be made obvious with var ious cues. The cues are then slowly eliminated in succeedin g items until the student is able to answer questions about the subject without the benefit of cues. Most pro g rams require written responses. The r eader could simply for mulate the response in his mind, but many programmers believe that the overt /

PAGE 47

Deo. 1962 CBBMIOAL EIGIIEERIIG EDUCATIOI 44 action of writing the response gives a more active role to the reader; and ( hence he learns better. .~no ther important use of written responses is in \. correcting the program. If an item in the program is too difficult, many students will answer it wrong. The programmer can then check these responses and modify the item to assure correct responses: Not all programmers believe that incorrect responses are necessarily bad. They may be advantageous if they can be used as a sign that the student needs more training on the subject of the question. The additional training can be added to the program by "branching". Depending upon his answer, the reader is told to go to one of several following items. If his answer is correct, he goes on to a more a dvanced itemIf he answers incorrectly, he goes on to a sequence which will correct his mistaken ideas. There may be more than one branch from a given item. Branches may also be used to skip items when a quick learner demonstrates a s~perior g rasp of the material. The other branch then includes more prac tice for th e slower learner. Branching is used in many other situations where the pro g ra~mer wishes to offer more th an one alternative sequence of items A 1:?ra~~~ m aJ some:tim~.~ b ~ t~ ken .a t ~!"le option o f the reader, if he feels he needs the additional knowledge in the branch. Branches may be of equal diffi cul ty, where the choice between them is based on an opinion of the reader, which is neither right or wrong. Figures, graphs, and lon g q uot a tions may be included in a program. They are often placed on separate pa g es a nd the student is asked to refer to them when he comes to a g iven item. There is no limit to the length of such additional material. It wou.ld be possible to include an entire art e le or book, and then ask detailed questions about the selection using the programmed items. In this way the te ache r can test the student's under standing of what he has re ad Devices for Presenting Programs: The program itself' is the core of proe;ra1nrned instruction_ The mechan ical de vic es for presenting the pro g rB rn 8. nd necessary supplernentar y material a re re a lly secondary. Howe v er the various de v ices a re sometimes suitable for different types of programming, so t hey wi.11 be discussed briefly. A good survey of these devices is g iven in Reference (A-12). Teachin 6 Machines are mechanical de vices for presenting one item at a time to the student. They usually require that the student make his re sponse (by writing the answer, pressing buttons for multiple-choice questions, etc.) before the correct answer a nd the next item appear. Often the ms chine is designed to make it impossible for the student to change his answer once he has seen t he correct answer Teaching machines range from a simple metal box costing a few dol la rs to complex devices including motion picture or slide projectors costing several hundred dollars Programmed Textbooks are essentia lly ~re several methods of pro g rammlng texts. different from El conventiona l textboo){, 11 paper tea,cl1 ing machines 11 There The resulting books look radically In the hori zont a lly-pro gramme
PAGE 48

4$ CHEMICAL EHGIIEERIBG EDUCATION Dec. 1962 running from top to bottom. The ree der starts at the top and reads to t h e bottom. Obviously, the answers to the items are visible on the same page, so some sort of shielding device is needed. This can be simply a sheet of pap~r. : that is lowered on the page as each item is answered; or it may be a special plastic cover that serves the same function. The vertical arrange ment permits a once-throu gh reading of the book; but it requires some sort of shield. A particularly intriguing programmed book is the scrambled text. Here the reader starts on the first page, where there is a discussion fol!owed by a multiple-choice question with answers. Depending on the answer he ch oo ses, the reader 1s told to turn to a specific page (never page 2). If he chooses the correct answer, the page he turns to tells him he is correct, and pre sents more material and another question. If he chooses a wron g answer, the page he turns to tells why his answer is wrong and gives remedial work, per haps extending through several more questions and branches until it finally returns to the main {correct) branch of the program. There may be several wrong answers in the multiple-choice question, each with its own remedi a l branch. Obviously such programs may become quite involved, with multiple branching. Reference (B-5) is an example of a scrambled text. Generally, programmed texts are less expensive than programmed teaching machines, because the former doesntt involve expensive mechanical equipment. For example, a proerammed text on elementary electronics costs $26 .25. A machine program for the identical material costs $ 70, and the necessary teaching m~chine costs $700 (Ref. A -3). Texts are more easily adaptable to branched programs. Com~uter-controlled teaching machines offer great promise for the fu ture (A-). Such devices could "tailor-make" programs, taking into account individual student differences in learning rate, educational backgroun d and aptitude. The computer could be responsive to e ach student's needs while it is handling a large number of students. At each step in the pro g ram, the com puter may modify the remaining pro g ram by considering such factors as 1. Promptness and correctness of the studentts answer. 2 Specific errors in t he answer 3. DR t a on the stud ent' s previous le a rning habits; such as readin g rate. 4. Personal data, including intelligence, special aptitudes, sex, etc. 5. Nature of the materi a l being studied. 6 Level of student motivation. Recent developments in computer-controlled teaching machines ar e d is cussed in Ref {A-4). An interestin g example of the computer-controlled te a chin g macrine is Plato II,(Programmed Logic for Au tomatic Teachin g Operations) developed a t the University of IllTnois. Reading material, figures, and questions a re presented on a television screen. The student types his answers usin g an electric typewriter. The a nswer may be in numbers, a lgebraic expressions, words, or sentences. When the student completes his answer he presses the "Ju dge" button and the computer ju dg es the correctness of the answer, flash. ing 11 01 ( 11 or 11 1' T o" beside the answer on the screen. If the a nswer is wrong, the student may e sk for additional help b~r pressin g the "Help" button. He then follows the computer which selects easier related material until the student indica.tes h e under standsby pressing the "Aha" button. He then returns to the question missed. Plate II has been used to teach mathematics and French. At present it is set up to handle two students simultaneously, but plans call for a larger number. The number of steps in the "Help" sequences is somewhat limited. Computer teachin g may eventually handle a large class, using only one computer for storage of the program and control. Each student could proceed at his own rate with his own television screen and typewriter. The cost of such an installation would be very high. Writing a teaching program for such a complex operation would be extremely difficu lt. The goal would be a pro gram which would anticipate every possible student error and would inc lude sufficient corrective material for even the slowest student. The result would be a machine which acted very muph like a human tutor, in that it would be al most completely adaptable to the needs of any specific student. Programming in College Courses Of the 122 programs commercially available in the fall of 19 62 only ten are appropriate for college classroom use (A-3). These include two ele mentary courses in statistics, two in general psychology, and one each in set theory, vectors, probability, chemistry, physics, and "Fortran" computer programming. In addition another half-dozen programs of post-high school level are listed. These include a series of short electronics programs pub lished by Varian Associates co verin g capacitors, klystrons, relays, and switches; one on math ematical logic, and one on basic electronics.

PAGE 49

De. 1962 OBRNICAL EIGilEBRI5G BDUCA!IOB The:re are ~ertainly more than ten college-level programs in existence, but many have been locally-dev e loped and are not yet commercially available. An excellent survey of progralllIT1,iI?-g activities pres ently under way in engine ~ring corses is given in Reference A-1. The .. survey shows considera ble interest in programmed instruction in engineering courses. Include d is a list of nearly 100 individuals preparing programs 0f inte rest to engineers ; but most of these are in the early stages of development. Programs in prep ar ation include eleme P. : tary courses in chemistry, physics, mathematics, sta tistics, mechanics, drawing, electrical engineering., computer prog:t'amming, an d a few others~ Several programs are being prepared in advanced mathemat ical sub jec ts, such as matrix theory, Boolean ale:ebra, Le place transforms, and vector analysis Many of these programs cover a limited subject which would be a small part of the full course. Undoubtedly some of these programs will be suf ficiently well~d~veloped and tested so that they can prove to be of wide spread use in engineering collegesThe survey indicates that many engi~ eering co lleges w ould use programs if they :were available. The in terest c~n ters on l!}athematics, engineering mechanics, and elebtrical engineering, but very little interest was showri in chernic ai engineering. Programming of ~ngineering subject s is bei ng encouraged by the com.mi ttee on Programmed J;nstruction of the ) ,.merican society for En gineeripg Educe tion. They have a nnounced plans for a workshop in programming tecl1niques for engineering faculty to be held in the swnmer of 1963 (A .. 1) To be appropriate for proe;ram..rning, a college course rnust huve a clearlydefined objective which can be stated in terms of overt behavior of the stu dent. \le 11-struc tu red subjects such as elementary ma the ma tics. a nd na tura l science seem to have well-defined objectives. On the other hJ3 nd, advanced courses in engineering design do not have specifically-defined obj ectives and could not be progrRmmed .. It is not necessary to program an entire course. Only those psrts which a re t-J ell -~truc tt~~ed need be programmed. ~he teacher can u se progr~ ms along with tradi tiona.l teaching techniques. For exa.mple, programmed Fortran in.;. struction could be inserted in the begin~ing of an eng~neering design course which required solution of design problems on the computer. Programrning, even wi th branpl1.ing, seems to r orce the student into a highly-structured patter,n. In a9vanced engineerlng courses emphasizing i dependent thought such structuring often would be undesirable. Engineering has emphasized the appi1cation of physical principles to the solution of complex problems. Al though the student may le~.rn the physi c?-1 principles by programs; it appears that the teachine of t11e sc:>l utl on of complex engineering problems is often too un$tructured to be appropriate for programrnine;. \/hat courses in a typical chemical engine~ring curriculum could be at lea st parti ally pfogrammed? A ny attempt to answer this que stion is, of neces si ~y, pure specul.ati on. Elernenta ry thermodynamics would appear to be suf ficient ly. deflned to permit progra1nming. Some of tb.e be sic c oncepts a.nd qefini tfons of m~ss and energy balances might be progre.mmed al th0ugh the more complex ba~ances could not a nd should not be. !'1any of the be sic con cepts of stage and .ra.te operations co uld be programmed, bu t it m1c).1t be dif f icul t to integrate t e he pro grams with non-programmed material Rn(l homework. Simi ls.r+y, basj_c concepts in ki~etics and pr ocess dyna1nics might be programme d although mo~t of these courses would not be. In ba.sic courses which are prerequisite to chemice.l engineering, pro grammed instruction appears to have many applicati : ons. ?-iost of freshman mathematics ; chemistry, and ph ysics could be programmed. Reference (B-6) is a c ollege physics program cons isting of 12~000 i terns. Ari experimente.1 fresh man chemistry prog~ am is given in Reference ( B-lQ).. Basic engineering mec han ics ean ,. be progranun:ed~ (An experime : ntal progra:rn in Kinematics 1s given in Refe re nce B-7) Elementary organic and ph:y sica : 1 chemistry appear to be a.p propri~te :for ~ !)me degree of programmi ng. An excellent pr~gra m for basic in struction on the slide rule is available in experi:mentel form ( B-9) A survey of several of the major publishers of engineering books indi.cates considerable inter e st in programmed colleg e texts, btit there are very few definite announcements of programmed engineering books to be published in 1963 o r 19 64. No programmed chemica 1 engineering texts have been announced. The problem would appear to be one of findin g authors sufficiently familiar with engineering and with programming. \ ..

PAGE 50

' 4,7 OBEM[CAL DGlBEERIHG EDtJCATIOI Dec. 1962 Interested engfneering teachers might try writing a program for a sec\ tion of one of their courses to evaluate the utility of programmed instruc tion. Guides on programming techniques are available (for example Ref. A-8 A-lo) Unfortunately, the techniques of programming are not sufficiently developed to make it either easy or enjoyable conclusions Most effort on programmed instruction has been devoted to hi gh school courses. AS a result, it is impossible to state whether programmed in struction can be successfully applied to college engineering courses. No conclusions can be drawn until extensive experimental studies have been made in engineering courses. In addition, much more research is needed to clarify the psychological principles of programmed instruction, so that the pr a ctice can be .put on a sound theoretical basis. .. Programming is still an art. Although the various schools of program. mers can cite psychological principles which they believe support their meth ods, !'ew parametric studie s to investigate their claims have been made. Gr e ater undeistanding of programinin g methods should preceae any ma. jor effort to program engineering courses. Methods developed for high school use are not necessarily the most appropriate for college courses. Some educational psychologists have questioned the criterion that requires most student re sponses to be correct. Perhaps more error would be desirable. One study (A-6) indicates that the activity of filling in blanks or writing out the answer may not be necessary. Possibly the value of a program is in the care fully-developed sequential arrangement of the important concepts to be learned. Most of the studies to date have compared independent programmed learn ing with i nstruction using conventional te.Jds and classroom discussion. Pos sibly studies using both programmed textbooks and classroom discussion would show that the combination is superior to either above. There is no need to substitute programmed texts for other techniques of teaching. In engineering courses particularly, a combination might be quite effective. A programmed text could be substituted for a conventional text in, for example, an existing thermodynamics course. Class discussion and homework could continue as before. Can programmed learning be used in chemical engineering education? It is too early to tell. The final answer to this question can be obtained only with the active participation of engineering faculties. Engineering professors must work closely with educational psychologists in developing and testing suitable programs. Although such cooperation would be a novel exper ience for both groups, it could prove very beneficial to teaching in chemical engineering. The potential of programmed learning is neatly summarized by R. D. Patton (A-11); "The !3nfant terrible of the moment is the teaching machine. No one oan say for sure wfiat kind of adult it may grow up to be. Meanwhile, it has struck fear into the hearts o~ all but the bold and perhaps the fool ish. The fear is that it may subvert the teaching art into a slick kind of game-playing, or at best so attenuate the humane aspects of the teacher-stu dent relationship as to cause the loss of the contagious joy of discovery which has so often sustained the intellectual life in the past. Yet there is much routine learning which must precede and accompany discovery; and it ls quite probably that auto-instruction devices may perform this function as well or better than the teacher in the flesh. Fortunately, many of tpe older and saner heads in this area of instructi on are urging caution and avoidance of extravagant claims until more experience is gained and more research com pleted. If good sense can carry the day, there is a very real possibility that the self-teaching device, whether it is a true machine or a programmed textbook, may relieve the human teacher of much drudgeryand permit even more 1ntimate contact between the older and younger learner". REFERENCES A Books and Articles Giving Information on Progra~s, Programming, and Programmed Learning. Al A 2. A 3. American Society for Engineering Education, Committee on Programmed Instruction "A Report on Programmed Instruction" J. Engr. Educ. 53:117 (Oct. 1962) Bushnell, D. D. : "Computer -based Teaching Machines", J. Educ. Res. ,5: ,528 ( 1962) Center for Programed Instruction: "Programs, t62A Guide to Pro grammed Instructional M,:iterials" U.S. Government Printing Office,

PAGE 51

Deo. 1962 CAL BIGI Il10 EDUCA'l'IOJr 48 .. .', r' ~ ,.,1, a ., !i!! Washington ( i962) This book 11 ets all programs COIDm~rc~ .ally aviilable in Septembe~ 1962, 1b includes a description or each program and aam ple pages. The index lists several additional programs or interest to engineers. Cost, $1.So. A 4. AS. A 6. Coulson, John (ed.) "Prograsnmed Learning and Computer-based In struction", John Wiley and Sons (1962} Deterline, w. A., ''An Introduction to Programed Instruction" Pren tice-Hall, Englewood Cliffs, N.J. (1962) A general introduction 1n eluding an illustrative program. Paperbo':lJld, 131 pages. Feldhusen. J. F., and A. Birt: "A Study ot Nine Methods ot Presenting Programmed Learning Material", J. Educ. Rea. 5S:46l (1962) A 7. Lumsdaine, A. A~ and R. Glaser (ed.) "Teaching Machines and Pro grkrrnned Learning~, National Education Association, Washington (1960). Thia 724-page book ~ollecte the important articles in the ti~ld up to 1960. coat: $7.50 A 8. A 9. AlO. All. Al2. Lya1taght, J., and Williams: ''Guide for Prograomted Instl"Uetion", John Wiley and Sona, New York. To be published February 1963 Mager, R. F.: "Preparing Objectives tor Progrswned Instruction", Fearon Publiahera, San Francisco (1962) Markle, s.H., L.D.Eigen, and P.K.Komoak1 "A Program~d Primer on Programine" Center for Programmed I,natruction 365 West En4 Ave., New York ( 1961) An ea eily-read programmed text giving an 1ntro duct19n to principles of writing programs. Cost: $2.00 Patton, R.D.: ''1J.'each1ng'' J. Higher Educ. 33:277 (1962) stolurow, L .11.: ''Teaching by Machine U. s. Government Printing Ot tioe (1961) A comprehensive sur-wr of the field. It includes descriptions of various machines, a discussion of the principles and practice of programming., a summary of research findings, and an extensive bibliography. 173 pages. Coat: $0.65 A 13. stolurow, L.N.: "Impl1cat1ons or Current Research and Future Trends", J. Educ. Res. 5S:Sl9 (1962) -----B p, Selection or Cormnercially-Available Programmed Textbooks of In terest to Engineering Teachers .. .. . ... J Bl. Baai~ S~ateme, Inc.: "Vectoran Appleton-century-Crofts, New York High school or beginning college levelJ 49~ frames; $2,75 B 2. Colman, H.L., and c.P; Smallwood: ''An Auto-instructional Introduction to Fortran Programming" McGraw-Hill Book co., New York College Level; 1000 tra=es; $3.95 BJ. Evans, J.L., and L .H. Homme: "Introduotor7 Statlatica" TMI Grolier~ New York.Advance high school beginning college levelJ 1700 frames; e10.oo General Education atatt: "Probability Modela ~t Random Proceaaee" General Education, Inc. Cambridge, Maaa College level; 800 trameaJ -~-00 BS. Hughes, R.J.~ and P. Pipe: "Introduction to Electronics" Doubleday, New York. Intended tor ae 1r-1nstruction in eleetron1oa tor people with little mathemat1oa background. Not a college text, but an interesting exalllple of the scrambled textbook. $4.95. B 6. Joaeph, A., and D. Leah7: "Programed College Phyaios" John Wile1 and Sons, New York. College level, but at a low mathematical levelJ 12,000 tramea; $15.oo B 7. Leah7, D., "K1nemat1oa" Center tor Programed Instruction, New York (Experim~ntal Edition) B 8. McFadden, M. "set,, Relations, and Functions", McGraw-Hill Book co., New York. High school and college level; llSO frames; 13.50. B 9. Morley, P~ nslide Rule Operation" Center tor Programmed Inatl9Uct1on, New York (Experimental Edition) B 10. Young, J .A. "Selected Top1ca, Freshman College Chem1atry" Available trom author, Kings College, Wilkes-Barre, Penna; 5000 trames; $8.oo ,..

PAGE 52

, I 4 importan t publicat ions from McGr awr-Hi 11 1 MOMENTUM, HEAT, AND MASS TRANSF ER An outatand1ng new aoden 1ntrodu.ot1~n to ohem1oal engineering. The 1 autm.011 haft torth the tund-.m~ntal ot montum, heat_ and maaa tranater and ha-Ye applied Ueia to eng1n ~1ng, pztoblem,, partloulai-17 thoae 111 the prooe'aa 1n4uetry~ The aa:nner ot xpoa1 tion aakea tbia book eapeoi ~l'J aui table tor an uaderpaduate text 1n naJ. neer1ng. Lucid writing reaul~a in o oapr.hena1 Jet oono1ae oo~r.age ot ,be tunttamental ot omen~um, heat, ana ti-anater, v1 th a nuabel' ot appl1oatlona in engineering 1n general, ~d oheadoal eng1a rlq la patloular. A d1acua1on ot the unit optti-atioQa 1a 1nolude4; but ; the or4er of the 1ubjeot 111.tt.r baa been de~er 1 11ned more : b,the cttruoture or tbe tbAtOZ'J' th~ bf the ~ual cl1T1 a1oa into unit operation~ 2. CH E MtCAL ENGIN EERI NG CALCULATIONS ) BJ '9BS! I BBMLff, Ina ti ~ute or Teotmoloa, and BR~-A BI~BBR, Eaeo Rai-ob and b.g1neerlng Coapan7. HoGrBlll Ser1ea 1n Chelld.oal ,!!f5!u .. r~}1g. 441 t9.50 P1m4wata1 oonoepta, i-atb.e~ tban tohnolo17 and eap1:nc11a, are treaaed tbl'oupout. Engineering te obn1qu ,,.b graphloal metho

xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EI8VYYCC5_DGU7NV INGEST_TIME 2011-09-15T19:43:21Z PACKAGE AA00000383_00005
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EBLD9UL9B_T55IG2 INGEST_TIME 2011-08-30T19:27:59Z PACKAGE AA00000383_00005
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES