INT. J. PROD. RES., 1974, VOL. 12, NO.

1, 2 1 - 4 4

Computer aided facilities design : An international survey

JAMES M. MOORE*
A wide variety of computer programmes for solving facilities design problems have
been developed. Since the arrival of programmes like CORELAP and CRAFT in
the early aixties, a variety of aspects of this problem have been approached using
computer technology. Kngineers and architects appear to be the developers and
users of most of these programnnes.
This paper describes the results of a survey attempting to summarize this
activity both in Europe and North America where most of the work is being
conducted. Computers are being used for both data reduction and graphics
associated with the design of buildings which range from industrial plants to
hospitals and schools.
It is apparent that computers are being used as design aids although few
practitioners are willing to delegate the whole job to a computer at this time.
Botli time shared programmes and the more conventional batch processing are
summarized. Heuristics include algorithms which vary in philosophy. The
advantages and limitations to construction algorithms, improvement algorithms,
vector analysis and graph theory are discussed. A catalogue of these programmes
including sources for further information is included.

Introduction
Man has been designing the physical facilities around him over all recorded
history. The town facilities which were designed and built are described in the
ancient history of Greeee and the Roman Empire. More recently the term
' facilities design ' has come to describe the process of planning physical equip-
ment, human beings, and structures into an efficient system designed to furnish
services or manufactured goods. Systematic Layout Planning by Richard
Muther (1961) is perhaps the book most widely read by practitioners on the
topic of facilities design. He attempts to provide procedures with sufficient
structure permitting practical problems to be solved economically with a
systematic approach.
Facilities design, sometimes called plant layout, is one of the classical areas
in which the industrial engineer has chosen to operate. Apple (1950) and
Ireson (1952) were among the early authors describing the plant layout problem
at book length. Moore (1962) introduced the use of operations research in
solving layout problems. In recent years solutions to layout problems have
been supplemented by computer technology.

Interdisciplinary nature of problem

Because of his historical interest in the problem, the industrial engineer
(Anderson 1973, Hillier and Connors 1966, Jansari and Gupta 1969, Khalil
1973, Lee and Moore 1967, Moore 1971) has been in the forefront of developing
computer programmes in this area. Other disciplines have also examined the
problem from their perspectives, for example, architects (Johnson 1970, Lee
• Department of Industrial Engineering and Operations Research, Virginia Polytechnic
Institute, Blacksburg, Virginia, U.S.A.
Presented at the 2nd International Conference on Production Research {Copenhagen) August 1973

PubUshed by Taylor & Francis Ltd., 10-14 Maeklin Street,, London WC2B 5NF.
22 James M. Moore

1971, Mitchell 1970, Newman 1966, Stewart and Lee 1972) have made contri-
butions to the area. People working in the area of business administration
{Buffa et al. 1963, Love 1969, Ritzman 1972, Vollmann 1968) have also developed
some original eontributions. Scholars from building science (Krejcifik 1969,
Moucka 1967, Portlock and Whitehead 1970, Whitehead and Eldars 1964)
have made some contributions as well as civil engineers (Banna and Spillers
1972, Spillers and Weidlinger 1970). As computer technology has developed, a
limited number of authors from computer science (Edwards et al. 1971, Seehof
and Evans 1967) and from computer graphics (Banna and Spillers 1972,
Teicholz 1968, 1972) have become interested in the problem.
The economist for many years has been interested in the location of economic
activities. Their original interest comes from the choice of alternative sites for
economic activities or factories. More recently the term ' economic activities '
has been interpreted at a variety of levels. As a result, some of the early work
done by the economists can be applied to the facilities design problem. Bind-
schedler and Moore (1961) draws upon earlier work of the economist by using
level curves for facilities design.

International activity
Research and development in computer aided facilities design has clearly
been taking place in many countries. Whitehead and Eldars (1964), Eilon and
Deziel (1966) and Portlock and Whitehead (1970) have been working on it in
Great Britain while a sizeable number of American authors have been making
contributions (Anderson 1973, Buffa et al. 1964, Edwards et al. 1971, Hillier
and Connors 1966, Jansari and Gupta 1969, Johnson 1970, Lee and Moore 1967,
Moore 1971, Moore and Mariner 1963, Muther 1961, Ritzman 1972, Seehof and
Evans 1967, Spillers and Weidlinger 1970, Stewart et al. 1970, Vollmann d aZ.
1968, Ward et al. 1970, Wesolowsky and Love 1972). Krejcifik (1969) and
Moucka (1967) have developed solutions in Czeehoslovakia through the applica-
tion of graph theory, in Germany, Warnecke and ZoUer (1972) have been
interested in the problem. Gero has done work in Australia and Seppanen has
done work in Finland. Swiss contributions have been made by Bindschedler
and Moore (1961) and Scharlig (1971). Although the French do not seem to be
directly making contributions to the facilities design problem, they have made
the major contributions to the area of graph theory (Benli 1971, Kaufmann
1964) which is fast becoming a useful tool in facilities design.
It would appear that the facilities design problem is not only interdiscip-
linary but also international based upon the locations and backgrounds of
interested contributors. To learn more about this rapidly growing area of
activity, a questionnaire has been mailed to known authors of existing computer
programmes in facilities design. This questionnaire ia attached as Appendix A.
Entries in this questionnaire are a summary of all responses.
Ill-structured problems
Newell (1969) describes what he calls the ill-structured problem. The ill-
structured problem has some of all of the following properties :
(1) It cannot be described exclusively in numerical variables.
(2) The goal to be attained cannot be defined in terms of quantitative
objective function.
 Computer aided facilities design : an international survey 23

(3) Algorithms that permit the best solution to be found and stated in
numerical terms do not exist.
It would appear that many of the problems dealt with in operations research
are ill-structured problems.
Benli (1971) suggests that the facilities design problem falls into this
category. In facilities design, the analysis includes quantitative data and
qualitative information. , Both kinds of information are essential if an efficient
facilities design is to be developed. The quantitative data, e.g. the daily
production output of a machine tool, or the weekly number of trips a lift-truck
makes between two departments needs to be collected and utilized in the design
procedure. At the same time the qualitative information, e.g. worker safety
and morale, or the aesthetic appearance of the exterior of the plant, must be
considered in the successful plant design. These two essential parts to the
facilities design problem lead us to the conclusion that we are dealing with an
ill-structured problem.
Ill-structured problems are handled in a variety of ways.
(1) Total enumeration
(2) Computerized techniques
(3) Interactive programming
(4) Heuristic programming
(5) Intelligent machines
Total enumeration is the brute force approach where the computer is used
as a fast calculator and evaluates all possible alternatives. Most engineering
problems utilizing computers operate at the second level, computerized tech-
niques. This involves solving a problem either mathematically or by simula-
tion. It requires skill on the part of the engineer to relate the problem on an
analytical basis. Assumptions are frequently added by the problem solver
later on. Sensitivity analysis gives the problem solver information about how
far in error he can be without significantly damaging the solution, even though
it docs not change the general nature of utilizing computerized techniques.
Interactive programming permits the best talents of both man and machine
to be utilized. The computer is effective at manipulating the large amounts of
numerical data typical to a facilities design problem but the graphical informa-
tion typical of facilities design problems is painful for the computer to store,
understand and analyze. Man, on the other hand, can quickly grasp significant
amounts of information from graphical presentations, although he is poor at
handling large quantities of data. Interactive programming utilizes the best
talents of both man and machine ; man handles the graphical analysis while the
cornputer does the data processing. A limited amount of work utilizing
interactive programming applied to the facilities design problem (Banna and
Spillers 1972, Moore 1971, Stewart and Lee 1972) has been done.
Heuristic programming involves algorithms which have the characteristic
of reducing the amount of search required to find an acceptable solution. A
heuristic programme contains a set of internal decision rules capable of inter-
nally modifying the direction of the search. It is this ability to determine how
well the solution procedure is going that made heuristic programming parti-
cularly applicable to ill-structured problems. The combination of heuristic
programming and interactive programming appears to have significant promise
24 James M. Moore

in solving facilities design problems (Banna and Spillers 1972, Lee and Moore
1967, Moore 1971, Stewart and Lee 1972). Perhaps ultimately, intelligent
machine will be able to solve ill-structured problems without the aid of man's
intelligence.
Architects, engineers and layout planners
Historically the industrial engineer has been interested primarily in the
arrangement of machine tools within a manufacturing facility. More recently
he has become involved in the design of service facilities like hospitals, banks,
newspaper offices, big city skyscrapers, and numerous other non-manufacturing
facilities.
The architect on the other hand has traditionally been interested in both the
interior and exterior of a building design. Whereas the industrial engineer's
orientation has traditionally been from the perspective of economy and
efficiency, the architects' perspective is more apt to be aesthetic. Clearly there
is a substantial overlap in the areas of operations of these two disciplines.
Unfortunately there is relatively little exchange of information between the
disci[)lines via their publications which are more or less mutually exclusive.
This summary survey is a modest attempt to examine activities on both sides
of the fence.

Types of layout planning projects

Layout planning projects fall generally into one of four levels of difficulty.
These levels are :
(1) New plant construction
(2) New layout in an existing building
(3) Re-layout of a restricted area
(4) Adding a machine in an existing layout
In general the new plant construction is the most expensive, but this level
occurs least frequently. On the other hand, adding a new machine into an
existing layout is the least expensive of the layout planning projects and
conversely occurs with rather high frequency. Although there are similar
characteristics between them, the level of difficulty differs.
The computer is utilized in quite different ways depending upon the degree
of difficulty of the layout planning project. Moore and Mariner (1963) describes
a computer technique which is easy to use when adding a new machine into an
existing layout. Figure 1 shows the iso-machinist movement curves calculated
by machine to determine the optimal location of a tool crib in a missile factory.
This same technique is useless when the problem involved is a new plant
con.struction. At the same time, the computer programme written by Teicholz
(1972) which designs site plans is quite useful at the new construction level but
harmless at the fourth level of adding a new machine. Figure 2 depicts an
improved site prepared by his programme called GRASP. The layout planner
needs to select with care the computer programme he needs to assist him in his
analysis. None of the programmes are universal.
Vollmann (1966) classifies computer programmes for facilities design as
either (1) construction type or (2) improvement type. The construction
algorithms build or construct a solution from the raw data. The improvement
 Computer aided facilities design : an international survey 25

Fig. 1

X
-tao

4S6 [•48 137 134 136

4 32 133 122 119 121

408 119 108 105 107

3B-4

360

336

BEST LOCATION for tool crib is at 312

x = 12 and j/ = 312. At this point, the
total amount of travel between the 2BB
candidate area and the 63 machine bays
is at a minimum. The three least signi-
ficant figures have been dropped for 264
simplicity. Level curves are shown for
80, 90 and 100.
240

216

192

[6K

144

120

96 149

72 164 153 ISO 152

48 179 169 165 167

195 165 181 183

2A 48 72 96
26 James M. Moore

Fig. 2

PROCRRMME GRflSP
JflNURRI 13G3
E«1C IElCHm.1
SCtlLE I INCH c 52 FEET

Fig. 3

No. OF RESPONDENTS

CONSTRUCTION

a: IMPROVEMENT
G
Q:
GRAPH THEORY
O

OTHER
 Computer aided facilities design : an international survey 27

algorithms require a feasible solution as part of the input. The heuristic works
on this feasible solution and improves it until no further improvements can be
found. In our survey there were more than twice as many construction
algorithms described as improvement algorithms. This distribution can be seen
in fig. 3. CORELAP was the first construction algorithm while CRAFT was
the first improvement algorithm. The names of the construction programmes
included in the survey are :
CASS LAYADAPT
C0L02 LAYOPT
C0MP2 LAYOUT
COMSBUL LSP
CORELAP MUSTLAP2
DOMINO PLAN
GENOPT PLANET
IMAGE SISTLAP
KONUVER SUMI
Other respondents to the survey indicate that they identify their pro-
grammes with the improvement type of algorithms. The names of the
programmes classified as improvement type by their authors are :
CRAFT OFFICE
GRASP PREP
KONUVER Terminal Sampling Procedure
LAYADAPT
Response to the survey was high although not 100%. Authors of the
following programmes did not respond as of this writing or could not be found
and therefore are omitted from the survey results :
ALDEP (Seehof and Evans 1967)
FRAT (Khalil 1973)
Hillier-Connors (1966)
MAT (Edwards et al 1971)
RMA Comp I (Muther 1970)
In general, the construction type of algorithms are more suitable to the new
plant construction problem. The improvement algorithms are more helpful
when the plant layout problem involves putting a new layout into an existing
building. A fair number of j)rogrammes now exist examining both of these
problems. Relatively little work has been done on the rclayout of an existing
facility. It would be extremely useful to have computer aids to examine the
inefficiencies of existing layouts. Such a programme would need to examine
the cost of moving and installation, recommend what machines should be
moved, and where machines should be moved, and what machines should not
be moved. This would be an extremely useful programme. To the best of
this writer's knowledge, no such programme exists.
Some specialized work has been done in the area of office layout like the
programme by Vollmann et al. (1968) which he calls OFFICE. Jansari and
Gupta (1969) also examines the specialized layout problem involved in office
design.
28 James M. Moore

RELationship chart input

Question number 5 in the survey reads ' Is the RELationship chart type of
data required as input ? ' Out of 21 responses to this question, 17 answered yes
and 9 answered no. The histogram in fig. 4 depicts this. The intent of the
question was to inquire if a matrix of input defining the priority of closeness
relationships between all pairs of activities was required. Of the seven negative
responses the author suspects in three cases that the respondent gave another
interpretation to the question. As is true of many questionnaires, the question
may have been confusing. It is quite possible by using the term ' Relationship
Chart ' in question ^, some respondents interpreted the question to mean the
adoption of Muther's six-step priority closeness rating scale (Muther 1961). It
was not the intent of the question in the survey to imply this. In a problem
where more than one activity is to be located, it is impossible for this writer to
conceive of a computer working on the problem without this type of data as
input, either in discrete or continuous form.

Fig. 4

RELATIONSHIP CHART DATA REQUIRED

10

YES

SURVEY AUTHORS
OPINION

Question 6 found 16 out of 21 respondents indicating their relationship

chart data is discrete. Here too it is possible that there was some confusion
about the question. Discrete implied to the writer a closeness rating scale like,
e.g. Muther's which involves six levels, while continuous suggested to the writer
a scale in which any whole number could be inserted. Some responses to this
question suggest confusion about the intent of the question, although there is
reason to believe that the responses are reasonably representative.
Question 15 reads ' Does your logic handle X (or undesirable) ratings where
activities must be a considerable distance apart ?' Twelve out of 26 responses
were positive. There is reason to believe that this proportion of positive
responses is misleading based on the comments to the follow-up question.
Question 16. The intent of the question was to inquire if the author has a
special sub-routine that checks the closeness of pairs of activities with X
ratings and takes some overt action in the heuristic to insure that such a pair
of activities are in truth some distance apart.
To the best of the writer's knowledge there are only two programmes which
do this. One is Muther's RMA Comp I (Muther and McPherson 1970) and the
other is Johnson's IMAGE (Johnson 1970). Although many of the other
 Computer aided facilities design : an international survey 29

programmes will accept X ratings as legitimate input data, among the con-
struction algorithms if a very small weighting is given to X I'ating or even a
negative rating that value of the relationship chart is merely ignored. There
is typically no overt sub-routines to check X ratings. In many of the improve-
ment algorithms the X ratings are merely deducted from the positive ratings in
moat of the scoring mechanisms. A typical response to question 16 involves a
comment about negative relationship chart values or nuisance values. In the
case of construction algorithms, this does not suggest a specific sub-routine
whose mission is to separate those pairs of activities with X ratings. Perhaps
question 16 was poorly stated, however, it is the author's opinion that responses
are misleading. We suspect that considerably fewer programmes actually do
something overt about X ratings than the survey suggests.

Design constraints
Every practical problem has its limitations and constraints. These design
constraints vary from problem to problem. It is quite likely that there will
never be a computer programme capable of dealing with all possible constraints.
There are, however, some common constraints which can be handled by some
of the programmes today. Questions 7, S and 9 in the survey deal with the
capability of programmes to resolve some of the more common constraints.
The single question in the survey with a uniform answer was question 7.
All 25 programmes are capable of handling any area unit, that is, it makes no
difference whether you plan in the metric or English system, nor does it make
any difference what order of magnitude of area unit is utilized. Ten res-
pondents indicated that the building outline is not required by their programmes.
Five indicated that the building outline is required as input with their pro-
grammes. The latter were primarily from the improvement algorithm group.
The RUGR algorithm from Czechoslovakia (Krejcifik 1969) based on graph
theory requires as input the fixing of one dimension of a rectangular building
outline. The other dimension is determined by the algorithm. If the length
of the building is fixed by the analyst and used as input, then the algorithm will
determine its width.
It is interesting to note that 24 respondents out of 29 answering the question
indicated that activities can be fixed in the layout. The early improvement
programmes had this capability whereas the first construction algorithms did
not. It now appears that both construction and improvement type of
algorithms have the capability to fix the location of activities depending on the
limitations of the problem.
Kach layout problem has unique limitations and constraints. It is im-
possible for any computer to consider all possibilities so that the layout planner
must step in to the solution procedure at this point. He may do this on an
interactive basis with his computer if it has that capability, or he may merely
evaluate the output from each computer run keeping in mind the practical
limitations of his specific problem.

Approaches in machine logic

The philosophy of the logic of programmes in the survey is summarized in
the histogram of fig. 3.
30 James M. Moore

Questions 13 and 14 in the survey were designed to obtain some descriptive

material very briefly about the heuristic. Question 13 is worded in such a way
that it makes more sense when discussing the construction algorithms than it
does the improvement algorithms. By far the most popular criteria xised for
determining the entry into solution might be called the use of the total closeness
rating (TCR).
TCR is the sum of the weighted closeness ratings tending to tug a given
activity in a variety of directions. It is analogous to the free body diagram of
mechanics. The bottle neck department is the department with the largest
number of ratings tending to pull it in a variety of directions. 8even of the
programmes utilized this method. It is used for entering activities into the
solution. This criteria is frequently used in manual solutions to layout prob-
lems. Its use in computer solutions is therefore understandable even though
it may not be the most efficient criteria from the point of view of computer
time.
LSP, CASS and the Tertninal Sampling Procedure all use the bias sampling
procedure. The bias sampling philosophy turns out to be very efficient from
the point of view of the computer which can generate random numbers very
rapidly. Randomly generated solutions are biased by the problems input data
in a modification of the heuristic originally proposed by Seehof and Evans
(1967) in their programme ALDEP.
RUGR from Czechoslovakia uses tbe characteristic of planarity from graph
theory as the basis of its heuristic. COMPROPLAN builds a three dimensional
relationship diagram rotating it in three dimensions until an acceptable two
dimensional view is found. Most of the respondents describing improvement
programmes indicate that the initial solution determines the entry of tbe
activities into tbe solution. It should be noted, however, that question 13 is
somewhat inappropriate in its wording when tbe discussion involves improve-
ment algorithms.
Ritzman (1972) indicates that his programme is based on tbe elimination
process. He describes it as eliminating ' undesirable center-location assign-
ments until it can go no further without ruling out all feasible solutions '.
Warneeke, in his survey response describes his two programmes, MUSTLAP 12
and SISTLAP 11, as being based on Vogel's approximation. Because the
author was unable to provide more details in the brief space allowed on the
survey, this writer can only assume that be is referring to Vogel's approximation
associated witb the transportation problem.
Question 14 involves bow activities are placed by the heuristic. The
responses suggest a wide variety of solutions to this problem. It involves the
area requirements and the shape of these areas. From the practical point of
view, departmental shapes are important and are usually based on a wide
variety of constraints. The following responses to question 14 aie associated
with tbe name of the computer programme :
see facing page
It is fairly obvious that there are substantial differences in the design of
heuristics capable of locating activities in a practical way. All of these
heuristics need to be improved in flexibility in order to cope witb the variances
on constraints.
 Computer aided facilities design : an international survey

Name of programme Typical responses of question 14

CASS, PLAN Shape always a square
PLANET Search perimeter for best location—the place as ' square '
as possible
LAYOUT Based on optimum position of first element in relation to
those already down
COMPROPLAN Replaces points by an area
COMSBUL Find adjacent location availability
RUGR Shape is determined in an iterative way, using the theorem
of flow through networks
CORELAP 8 Rectangular with specified area and modified shape is
moved over layout to find best fit

Evaluation capability
Of the 28 responses to this question, 22 indicated that their programme does
have evaluation capability. All of the improvement programmes have the
capability to evaluate their solutions which is required by the nature of the
algorithm. In addition, seven of the construction type algorithms have the
capability of evaluating their own solutions. The criteria used for evaluations
seemed to break into two categories. The first involve total handling costs,
while the second was the sum of weighted distanees. The weights on the
distances are usually the weights presented in the relationship chart input.
Respondents were approximately equally split between these two criteria of
evaluation.
Some programmes like CRAFT and ALDEP output the evaluation score
in the form of a single number which is helpful in comparing alternatives, but
not much help in making manual adjustments to a given solution. Other
programmes like CORELAP 8 provide a detailed evaluation for each pair of
activities. This feature highlights immediately the trouble spots in the com-
puter solution so that the layout planner can make appropriate adjustments.
Although this feature nearly doubles the computer time of CORELAP 8, it
performs a large data processing task which few planners have the patience to
do by hand.

Computational characteristics
Three quarters of the programmes included in the survey were written in the
FORTRAN language. Three of the programmes have been written in ALGOL
and one each in Minsk, BASIC, PL/1 and APL. Clearly FORTRAN must be
more easily accessible to layout planners. See fig. 5 for histogram.
Questions 11 and 12 in the survey dealt with upper limits to computer and
programme capacities. In response to the question, ' What is the maximum
nuTiibcr of activities allowed ?' responses ran from a minimum of 30 to infinity.
I.J.P.R. J,
32 James M. Moore

Fig. 5
COMPUTER LANGUAGE
NUMBER OF RESPONSES

FORTRAN

ALGOL

BASIC

APL

The usefulness of a programme capable of handling an infinite number of pro-

grammes would appear to be somewhat marginal in the practical sense. A
more typical response to the maximum number of departments or activities
was around 99. Responses to question number 12 indicated the maximum
number of area units or elements ranged from 100 to infinity. A typical size
appeared to be a matrix about 30 x 30. Responses to both of these questions
indicated that the questions may have been confusing.

Output format
As the histogram in fig. 6 shows, the printed matrix is by far the most
popular output format. This is true because even a minimal digital computer
has printer capability. Only fancier machines have plotters and cathode ray
tube displays. In our survey five programmes were found to have the capa-
bility of displaying on a cathode ray tube. RUGR, the algorithm based on
graph theory, depicts a planar graph equivalent to the layout problem in the
format of a relationship diagram. This appears to be the only programme in
the survey with this capability. Figure 7 provides an illustration of this unique
capability. Note that line widths represent the closeness priority in the
diagram.

Fig. 6
No. OF RESPONSES
 Computer aided facilities design : an international survey 33

Fig. 7

Courtesy of Jan Moucka.

Interactive programmes
Although most of the computer programmes are designed for batch pro-
cessing, eight respondents indicated that their programmes have interactive
capability. This capability permits the user to interrupt during execution of
the programme to assist the heuristic with counsel and adviee. It is these
interactive programmes which permit the complementary talents of man and
the computer to jointly assist one another.
Interactive CORELAP (Moore 1971) uses standard teletype terminals on
time-shared systems for maximum availability. It operates in the U.S.A. on
the General Electric Time-Shared System. Arrangements are being made for
it to operate on an international scale through the General Electric Time
Shared System using satellite transmission. Figure 8 depicts the global
capability currently in operation involving three continents. COMPROPLAN
(Stewart and Lee 1972) operates on the PDP-15 computer system of the
Digital Equipment Corporation. It includes a lite pen, a graphics console and
a 17 or 21 inch screen. Figure 9 shows the computer graphics capability of the
PDP-15 system. The upper portion of the figure was drawn by the computer
and can be compared with the photograph of the actual building designed by
Perry, Dean and Stewart, Architects of Boston.
These two programmes perhaps span the spectrum of user assisted pro-
grammes. Interactive COREJ^AP utilizes a general purpose time-shared
system with minimal special hardware required while COMPROPLAN repre-
sents a captive special purpose computer with high powered graphics capability.
Needless to say the costs rise sharply as equipment demanded becomes more
special purpose.

Diagnostics
The histogram in fig. 10 depicts the degree to which error diagnostics are
contained in the various programmes. An equal number of respondents said
their error diagnostics were fair as said their programmes had none. Only five
C2
34 James M. Moore

respondents out of 27 felt their error diagnostics were good. The reader should
keep in mind that the respondents were all people who have worked on the
development of their respective programmes. It is reasonable to expect a slight
bias towards the optimistic end of the scale from tliis particular group. A user
involved in running someone else's programme very frequently initiates errors
undreamed of by the author of the programme. One concludes that if authors
themselves admit that their diagnostics are poor, then the casual user of
someone else's programme had better be forewarned that he should anticipate
difficulties.
In response to the inquiry about what types of errors were included, most
respondents indicated that error statements cover input data and parameters.
A limited number of programmes have format error statements. An even
smaller number of programmes have error statements explaining fatal and non-
fatal problems whieh the heuristic is unable to handle. Figure 11 summarizes
the responses to this question. It would appear that a great deal of work has
yet to be done in the diagnostics area if facilities design computer programmes
are to be widely utilized.
 Computer aided facilities design : an international survey

Fig. 9

f .• T' ''

.„ J,, J -i; r. TIFFF-f i J 4

IT zz
i j;
I . J
•]i ....
n
-i , — J ; j ... ^
'tJ
' L! j Uli .U LH
I.
-1
—1.
•td-b

Fig. 10
FREQUENCY OF RESPONSES
5

NONE

LIMITED
ERROR

FAIR
DIAGNOSTICS

GOOD
36 James M. Moore

Fig. 11
FREQUENCY OF RESPONSES
5

INPUT DATA

TYPE OF
ERRORS PARAMETERS
COVERED

FORMAT

How to use computers in layout planning

The typical layout planning project goes through four phases. They are :
(1) Relationships
(2) Space
(3) Adjustment
(4) Evaluation
Of course these phases can be executed manually, however, the numerous
computer programmes now available can be of assistance during some of
these phases. Most of the computer programmes described in this paper require
the layout planner to define his relationships and his space requirements.
Given this input information, the computer is able to design a layout. Very
little work has been done using computers to help the layout planner develop
his relationship data. Two notable exceptions are Ardini (1972) and Seehof
and Evans (1967). Ardini has prepared a programme to cumulate volume-
variety information (sometimes called Product-Quantity PQ) and prints a
histogram curve which is helpful to the layout planner. Seehof describes a
programme for determining inventories of machines and furniture which can be
used as input to his ALDEP.
The adjustment phase at this time must be done by the layout planner.
None of the computer programmes have sufficient flexibility to consider any-
thing but an extremely limited number of practical limitations or modifying
considerations. These limitations are exclusively unique to the specific project.
The evaluation yardsticks currently used by computer programmes are
quite limited. The typical decision maker has many more yardsticks he
wishes to apply to a layout solution. At this point in time the computer is
only able to use quantitative criteria for evaluation. It is clear that qualitative
criteria are used by most decision-makers in approving final layout plans.
The final question of the survey asked if the respondent thought tlie
programmes currently available for facilities design were 'extremely useful,
marginally useful, useless ' or ' not only useless but expensive as well '. One
should keep in mind that all respondents have devoted some if not considerable
time to developing these programmes. This survey is not a random sample.
As fig. 12 shows this group is a little biased. It would be interesting to survey
practicing layout planners and ask a similar question.
 Computer aided facilities design : an international survey 37

For the benefit of the reader, Appendix B includes the names and addresses
of respondents to the survey who indicated that their computer programmes are
available for general purpose use. Interested users can contact the authors
directly.

Fig. 12
NUMBER OF RESPONSES
5

EXTREMELY USEFUL

MARGINALLY USEFUL

USELESS

BOTH USELESS + EXPENSIVE

A word of caution : None of the current computer programmes are suffi-

ciently refined to be used thoughtlessly. Successful users of computer pro-
grammes in the facilities design area have used them with moderation and good
judgement. Computer solutions have frequently been drawn upon to develop
additional alternatives to manually prepared solutions. One layout planner
working for an internationally known computer company has suggested to this
writer his apprehensions about computer solutions to layout problems. He
found that his company managerial decision-makers were willing, even anxious
to accept any layout solution prepared by computer. He found that computer-
proposed layouts were much more saleable than manually prepared layouts
even though their quality was substantially lower.

Summary
The computer aided facilities design activity is the main topic of this paper
in which the results of an international, inter disciplinary survey are sum-
marized. The problem is described as fitting tho ill-structured form of problem
well known in operations research. The relationships between architects,
engineers and layout planners is discussed along with a classification of layout
planning projects into four categories ; namely, new construction, new layouts
in existing buildings, relayouts and adding of single machines.
The survey shows a wide varietj'^ of approaches using the computer as an
aid. The two prevalent approaches seem to be the construction algorithms
and the improvement algorithms. Graph theory and linear programming are
invoked to a limited degree. Only a very limited number of design constraints
are built into existing programmes at this time. The layout planner must still
do almost all of the adjustments and should <;ertainly check the evaluation
done by the computer. Output is available from printers, plotters and tube
displays, but the printer is by far the most widely used at this time.
38 Jam,es M. Moore

Work needs to be done to develop programmes which will prepare RELation-

ship Charts and space requirements. Much more work remains in improving
the computers ability to evaluate solutions with an eye towards the practical
problems.

Appendix A
SUMMARY OF
SURVEY OF COMPUTER AIDED FACILITIES DESIGN

1. Have you ever developed a computer programme for aiding in the design of
physical facilities, industrial or otherwise ?
28 Yes [Please continue) 3 No {Please skip to question 30)
2. What is the name of your programme ?
3. Please describe in 25 words or less what your programme does.
4. Is the programme of the
22 Construction type (starts with data only and constructs a solution from
scratch) ;
10 Improvement type (requires feasible solution as input and makes
progressive improvetnents until no further can be made) ;
1 Based on Graph Theory mathematics ;
2 Other. Please specify
5. Is the RELationship Chart lype of data required as input ?
Actual Results
1"? Yes 9 No (Please skip to Question 7)
26 0 Author's Opinion
6. / / yes, to the previous question, what class of data is acceptable in the RELation-
ship Chart ?
16 Discrete 5 Continuous
7. What measurement units are required ?
Square Feet
Square Meters
25 Any area unit
2 Other. (Please specify)
8. The building outline is 5 | required as input.
10 optional as input.
9 not required.
9. Can the location of activities be fixed in the layout ?
24 Yea 5 No
 Computer aided facilities design : an iiiternational survey 39

10. What computer language is used ? FORTRAN 21

ALGOL 4
Minsk 1
Other {Please specify)
BASIC, APL, PL I
] 1. What is the maximum number of activities allowed, i.e. maxim,um number of
activities or departments to be located in the layout ?
min = 30 ; max = oo

12. What is the maximum number of area units which can be accommodated in the
laijout matrix ? In other words, how big can the layout matrix be ?

] 3. How is the entry of activities into the solution deMrmined ?

7 TCR 2 Vogel's Approximation
1 Planar Graph 2 Initial Solution
1 3-Dimensional Optimal
3 Random Bias 1 Elimination Process

14. What method is used for placing activities into the layout, that is, how is their
location and shape determined ?

15. Does your logic handle X {or undesirable) ratings where activities must be a
considerable distance apart ?
12 Yes 14 No [Please skip to Question 17)
1 25

16. / / you answered yes in the previous question, how is the X rating handled ?
Small or negative REL values

17. The computer output is in the form of :

1 1 a printed list 5 a cathode ray tube display
22 a printed matrix 2 other. (Please specify)
8 a graphical plot

18. Is the programme user assisted, i.e. can the user interrupt during execution and
assist the heuristic ?
8 Yes 19 No

19. Can your programme handle multi-storey layouts ?

15 Yes No

20. If you answered yes to the previous question, how does it handle this difficult
problem, ?

21. Error diagnostics contained in the programme to help the user are
8 non-existant Input data 13
0 limited Format 9
8 fair Parameters 11
5 good. What types of errors are covered ? Other. {Please specify)
40 James M. Moore

22. Does your programme evaluate alternative layouts ?

20 Yes 7 No {Please skip to question 24)

23. / / yes to the previous question, what criteria of evaluation is utilized ?

S (Weighted) (Distance)—5 Total or Handling Cost—6

24. Can a solution which has been prepared manually be entered into your pro-
gramme and be evalvuted ?
21 Yes 7 No

25. When was your original work completed ? 61 CRAFT, 64 Whitehead,

64 CORELAP

26. When was it most recently modified ? 72

27. Is your programme available to other interested users ?

21 Yes No

28. / / yes to the previous question, whom should be contacted ?

Name :
Address :

29. / / yes to question 27, is there a charge and if so how much ?

30. Do you know anyo7ie else who has worked on or is working on the development
of cojnputer programmes for facilities planning ?
13 Yes 15 No {Skip to question 32)

31.7/ yes to previous question, they are

Name :
Address ;

32. Have you used any computer programmes in the desigyi of facilities ?
24 Yes 6 No {Skip to question 35)

33. What is the name of the programme you have used ?

34. What is your estimate of the usefulness of the programme ?

10 extremely useful
8 marginally useful
2 useless
not only useless but expensive as well.

35. What is your level of interest in this topic ?

30 high, please forward a copy of survey results.
enough to respond but don't bother sending results.
1 very low, please don't bother me again.
 Computer aided facilities design : an international survey 41

Appendix B
Sourees for programmes in survey whieh are obtainable

Name of
programme Name and address of author (or source)

C0MP2 Mr. Eric Teicholz Laboratory for Computer Graphics,

Graduate School of Design, Harvard
University, Cambridge, Massachu-
setts, U.S.A.
COMPROPLAN Mr. Kaiman Lee Perry, Dean and Stewart, 955 Park
COMSBUL Square Building, Boston, Massachu-
setts 02116, U.S.A.
Major Robert C. Lee Engineering Management Assoc, Room
C0RELAP8 590 UR, 360 Huntington Avenue,
Boston, Massachusetts 02115, U.S.A.
Dr. Gordon Armour Dr. Elwood Buffa, Graduate School of
CRAFT Management, U.C.L.A., Los Angeles,
Calif. 90024, U.S.A.
DOMINO Mr. Robert Dillon Welton Becket & Associates, Archi-
tects, Loa Angeles, California, U.S.A.
GRASP Mr. Eric Teicholz Laboratory for Computer Graphics,
Graduate School of Design, Harvard
University, Cambridge, Massachu-
setts, U.S.A.
IMAGE Dr. T. E. Johnson Department of Architecture, Massachu-
setts Institute of Technology, Cam-
bridge, Massachusetts, U.S.A.
Tnteractive Dr. James Moore Engineering Management Associates,
CORELAP Room 302A, Whittemore Hall, V.P.I.
& S.U., Blacksburg, VA 24061, U.S.A.
KONUVER Dr. Warnocke Institut fur Produktionstechnik und
Automatisierung (IPA) an der Uni-
versitat Stuttgart, 7000 Stuttgart 1
Holzgartenstr. 17, Postfach 951
LAYADAPT Dr. Warnecke Institut fur Produktionstechnik und
Automatisierung (IPA) an der Uni-
versitat Stuttgart, 7000 Stuttgart 1
Holzgartenstr. 17, Postfach 951
LAYOPT Mr. Raimo Matto Tmatran Voinia Oy, Box 10138,
Helsinki, Finland
LAYOUT Mr. John S. Gcro Dept. of Architectural Science, Univer-
sity of Sydney, Sydney, Austraha
Love* Dr. R. F. Love Graduate School of Business, 1155 01)-
*Authors name, pro- scrvatory Drive, University of Wis-
gramme lias no name consin, Madison, Wisconsin 53706,
U.S.A.
MUSTLAP2 Dr. Warnecke Institut fur Produktionstechnik und
Automatisierung (IPA) an der Uni-
versitat vStuttgart, 7(X)0 Stuttgart 1
Holzgartenstr. 17, Postfach 951
OFFICE Dr. Tliomas Vollman Management Science Department,
Ctjllege of Business Administration,
University of Rhode Island, Kings-
ton, Rhode Island, U.S.A.
42 James M. Moore

Appendix B^continued

Name of
programme Name and address of author (or source)

PLAN Mr. K. McRoberts Dept. of Industrial Engineering, 212

Marston Hall, Iowa State University,
Ames, Iowa 50010, U.S.A.
PLANET Mr. Michael P. Industrial & Systems Engineering,
Deisenroth Georgia Institute of Technology,
Atlanta, Georgia 30332, U.S.A.
PREP Mr. David Anderson 2933 E. Binkley # 4 1 , Dallas, Texas,
U.S.A.
RG&RR Mr. Jan Moucka Research Institute for Building and
Architecture, Prague, 7, Letcnska 3,
Czechoslovakia
Ritzman* Dr. L. P. Ritzman College of Administrative Science, Ohio
*Authors name, pro- State University, Columbus, Ohio
gramme has no name 43210, U.S.A.
SISTLAPM Dr. Warnecke Institnt fur Produktionstechnik und
Automatisierung (IPA) an der Uni-
versitat Stuttgart, 7000 Stuttgart 1
Holzgartenstr. 17, Postfaeh 951
SUMI Prof. Spillers Civil Engineering Department, Colum-
bia University, 116th & Broadway,
New York, New York 10025, U.S.A.
Terminal Mr. G. Hitehings Dept. of Meehanical Engineering and
Sampling Engineering Production, University
Procedure of Wales, Institute of Science and
Technology, King Edward VII
Avenue, Cardiff CFl 3NU, U.K,

On a d«'?veloppe une grande varietti de programmes d'ordinateur pour resoudre

(os probl^mos poses par la conoeption de centres de service. De puis la creation
de programmes comtne CORELAP et CRAFT au dobut dos annees 60, on a
considtire un grand nombre deaspects de ce problemo, en utilisant la technologie de
I'ordiiiatuer. II seiiibie que r;e soit les ingmieurs et les archit.&ct6S qui aient
dcveloppe et utilise la piupart de ces programmes.
Cet article decrlt les resultats d'line onquOte qui essaie de resumer cette activitu
en Europe et en Ainerique du Nord en mome temps, oil la plus grande partie des
recherches soiit effetituees. On se sert d'ordinateurs pour la rediiction des donnees
ainsi que pour los traces assocics a la coticoption de batiments allant d'installations
industrieiles a doa bopitaux et ecoles.
II est evident qu'on se sort des ordinateurs pour aider au niveau d© la conception
bienque peu d'operateura soient preta a deleguer toute l'operation a un ordinateur
a ce .stade-la. On resume les programmes a travail en siinultane et le traitement
par lots plus conventionnel. L'heuristique ooniprend des aigorithmes dont la
philostiphie varie. On discute les avaiit.ages et lea liinites des algoritbmos de
construction, des algoritbmes d'amelioration, de I'analyse de vectour et de la
theorie de trace. On inclut dan3 cet article un catalogue de ces programmes, y
compris les sources oil trouver tous renseignenients supplementaires.

Zur Losung von Konstruktionsproblemen ist eine gro^e AuswabI von Computer-
programinen ontwickelt worden. Seitdem Anfaiig der sochziger Jahre Programme
win CORELAP und CRAFT eingefiibrt wiirdeii, wind durcb Einsatz der Coinputer-
technologie verschierlene Gesicbtspunkte diosi>s Problems angepackt worden.
Ingenieure und Arcbitekten sind wabrscbeinlich fiir die Entwicklung und den
Gebraucb der meisten Programme vorantwortlich.
Diesel- liericht beschroibt die Ergebnisse einer Untersucbung, bei der versucbt
 Computer aided facilities design : an internuti(mal survey 43

wui-de, diese Tatigkeit sowohl in Europaalsauch in Nordamerika zuaammenzufasaen,

wo der gropte Teil dieser Arbeit durchgefuhrt wird. Computer werdeii sowohl fiir
die Dateiiverdichtung als auch graphische Betriebsarten benutzt, di© mit dem
Entwurf von Gebaudon verbunden siiul, bei denen ea sich iim industriolle Aniagen
bis zu Krankenhiiusorn und Sehulen handoln kann.
Es ist offenaiehtiich. dap Computer als Entwurfshilfen eingesetzt werdon,
obwDhl wenige Praktiker beroit sind, den ganzen Job gegonwartig einom Computer
7.U iiberlassen. Sowohl dio zoitlich verxahnte Verarbeitung mehreror Programme
als auch die herkommlichero Stapelverarbeitung werden zusammengefapt. Die
Heuristik umfa&t Algorithmon. deren Philuaophio veraadorlich ist. Der Bericht
enthalt eine Bespreehung der Vorteile und Begrenzuiigen von Konstruktion-
salgonthmen, Verbesserurig.^algorithmen, Vektoranalysis und Zoiclientheorie.
Er enthiilt ebeiiftUIs einon Katalog diewer Programme, einschlieplich der weiteren
Informationsquelksn.

REFERENCES
ANDERSON, D . M., 1973, New plant layout information system, Ind. Engng 5, 32
APPLE, J. M., 1950, Plant Layout and Materials Handling (New York : Ronkld Press)
ARDINI, J., 1972, Volume-Variety (VOLVAK), Unpublished paper. North-
eastern University. Boston, U.S.A.
AitMouR, G. C, and BUFFA, E . S., 1963, A heuristic algorithm and simulation ap-
proach to relative location of facilities, Mgmi Sci., 9, 294.
BANNA,S.AL,andSPiLLERS,W. R., (eds.). 1972, SOMI—An interactive graphics space
allocation system. Department of Graphics, Columbia University, New York.
BENLI, 0. S., 1971, A review of the approaches to the solution of some ill-structured
problems (unpublished paper), October.
BEROE, C , 1958, Theorie des Graphes et ses AppUcatio7is (Paris : Dunod)
BINDSCHEDLER, A., and MOORE, J. M., 1961, Optimal location of new machines
in existing plant layouts, J. Ind. Engng, XII, 41.
and HALE, M. E . , 1971, Modular allocation techni-
EDWARDS, H . K . , GILLETT, B . E . ,
que, (MAT), Mg7nt Sci., 17, 101.
EILON, S., and DEZIEL, D . P., 1966, Siting distribution center, an analogue computer
applications, Mgint Sci., 12, B245.
HILLIER, F . C , and CONNORS, M. M., 1966, Quadratic assignment problem algor-
ithms and the location of indivisible facilities, Mgmt Sci., 13, 42
IRESON, VV. GRANT, 1952, Factory Planning and Plant Layout (Englewood Cliffs
New Jersey : Prentice-Hall, Inc.). '
JANSARI, J^, and GUPTA, I., 1969, A program for plotting plant layout, J. Ind. Engng,

JOHNSON, T . E . ,1970, IMAGE : An interactive graphics-based computer system

for muIti-constrained spatial synthesis, MI'I' Department of Architecture
September. '
KAUFMAN, A., 1964, Metkodes et Modeles de la Recherche Operationelh (Vol II) (Paris •
Dunod).
KHALIL, T . M., 1973, Facilities Relative Allocation Technique (FRATI //*( J
Prod. Res., 11, 183. ;-
, M., 1969, Computer-aided plant layout. Computer Aided Design 2,• 7 ••
1966, Matheinaticke Metody v Projektovem Rozhodovani (Mathematische
Methoden m Projektentscheidungen), Forschungsinstitut fur Bauwesen und
Architektur, Prague.
LEE, K 1971, Computer programs for architects and layout planners. Proceedings
of the American Institute of Industrial Engineers National Conference Boston
LEE, R . C , and MOORE, J. M., 1967, CORELAP—Computerized RELationship
LAyout Plannmg, J. Ind. Engng, XVIII, 195.
LEVIN, P. H., 1964, Use of graphs to decide the optimum layout of buildings Archi-
tects' J., 7 October, p. 908.
LOVE, R . F., 1969, Locating facilities in 3-dimensionaI space by convex programming
Nav. Res. Logist. Q., 16, 503 ; 1972, A cotnputational procedure for optimally
locating a facility with respect to several rectangular regions J reo Sci
12, 233. e> . y- .,
44 Computer aided facilities design : an international survey

MITCHELL, W . J., 1970, Notes on approaches to computer-aided space planning. School

of Architecture and Urban Planning, University of California, Los Angeles.
MooBE, J. M., 1962, Plant Layout and Design {New York : The Macmillan Company);
1971, Computer program evaluates plant layout alternatives, Ind. Engng, 3, 19.
MOORE, J. M., and MAUINER, M., 1963, Layout planning—new role for computers.
Mod. Mater. Ilandl, March.
MoufiKA, J., 1967, Methode des entwerfen von grundrissen eingeschossiger Bautcn
(Methods of design of layouts of single-storey buildings), / Vth International
Congress of Applied Mathematics, Weimar, June.
MUTHER, R . , 1961, Systematic Layout Planning (Boston: Industrial Edueation
Institute).
MUTHER, R . , and MCPHERSON, K . , 1970, Four approaches to computerized layout
planning, Ind. Engng, 2, No. 2, 39.
NEWELL, A., 1969, Heuristic programming : ill-structured problems. Progress tn
Operations Research, Aronofsky (ed.). Vol. I l l , (John Wiley), p. 363.
NEWMAN, W . M., 1966, An experimental program for architectural design, Comput. J.,
9, May.
PoRTLOCK, P. C, and WHITEHEAD, B., 1970, A programme for practical layout
planning. Build. Sci., 6, 213.
RITZMAN, L . P., 1972, The efficiency of computer algorithms for plant layout,
M'gmt Sci., 18, 240.
ScHARLio, A., 1971, La localisation optimale d'line entrepnsc. Revue Suisse
d'Economie Politique et de Statistique, 107^ Annee, Fascicule 3, p. 599 ; 1969,
Localisation Optimale et Theorie des Grapbes Cabiers Pareto No. 19 (Droz
ifiditeur, Geneve), 156 pages.
SEEHOF J. M., and EVANS, W . , 1967, Automated layout design program, J. Ind.
Engng, X V n i , 69.
SEEHOF, J . M., EVANS, W . , FRIKDRICHS, I. W., and QUIPLEY, I. J., 1966, Autoniated
facilities layout programs. Proceedings of the National Meeting, Association .
for Computing Machinery.
SFILLEBS, W . R . , and WEIDLINGER, P., 1970, An algorithm for space allocation,
Proceedings of the tith Annual Association for Computing Machinery Urban
Symposium, p. 142.
STEWART, —., TEICHOLZ, E , D . , and L E E , K . , 1970, Computer Architecture Programs.
Three Volume Reference Manual, Perry, Dean & Stewart, Architects, Boston,
Mass., June. .
STEWART —., and LEE, K . , 1972, COMPROSPACE : Interactive computer graphics
in'the real world, EDRA 3//l/?S Conference, Los Angeles, January ; 1971,
Can a 54-year old architectural firm find romance and happiness witb an
interactive computer system ?, Progve Archit., July.
TEICHOLZ, E . D., 1972, GRASP—Generation of random access site plans, un-
published paper. Laboratory for Computer Graphics, Harvard University,
Cambridge, Mass.; 1968, Architecture and the computer, Archit. Eorum,
September. .
VoLLMANK, T. E., NUGENT, C. E . , and ZARTLER, R . L., 1968, A computerized
model for office layout, J. Ind. Engng, XIX, 321.
VOLLMANN, T . E . , BUFFA, E . S., and ELWOOU, S., 1966, The facilities layout problem
in perspective, Mgmt Sci., 12, B-450.
WARD, W . S., GRANT, D . P., and CHAPMAN, A. J., 1970, A PL/1 program for arcbi-
tectural and space allocation. Proceedings of the 5th Anniuil Association for
Computing Machinery Urban Symposium, New York, August.
WESOLOWSKY, G. 0., and LOVE, R . F . , 1972, A non-linear approximation method for
solving a generalized rectangular distance Weber problem, Mgmt Sci., 18, 656.
WESOLOWSKY, G. O., 1971, The optimal location of new facilities using rectangular
distances. Ops Res., 19, 124.
WHITEHEAD, B . , and ELDARS, M . Z., 1964, An approach to the optimum layout of
single-storey buildings. Architects' J., 17 June, p. 1373.
WHITEHEAD, B . , and AGRAA, 0. M., 1968, Nuisaiiec restrictions in the planning of
singlc-storcy layouts. Build. Sci., 2, 291.
ZOLLER, K., and ADENDORFF, K . , 1972, Layout planning by computer simulation,
A.I.i.E. Trans., ^, 116.