Comput. & Indus. Engng Vol. 9, No. 2, pp. 141-148, 1985
Printed in the U.S.A.

0360-8352/85 $3.00 + .00 © 1985 Pergamon Press L td

FACILITIES LAYOUT A SURVEY OF SOLUTION PROCEDURES
REUVEN R. LEVARY Operations Research Center, M.I.T., Cambridge, MA 02139, U.S.A. and Department of Management Sciences, Saint Louis University, Saint Louis, MO 63108, U.S.A.

and
SYLVIA KALCHIK Monsanto World Headquarters, Saint Louis, MO 63167, U.S.A. (Received 26 January 1983; received for publication l June 1984) Al~traet--The main characteristics of the most-used solution procedures for the facility layout problem are summarized. The characteristics include input required, limitations, type of output obtained, and some other general characteristics. Quantitative and computer-based models will sometimes produce odd layout shapes. For this reason, optimization models and computer programs designed to plan facility layouts, based on optimizing certain objectives, can be useful only for determining some guidelines for the planners. While these models can be used for planning alternative layouts, considerations of qualitative and personnel factors should be given during the layout finalization phase. INTRODUCTION

One of the tasks in layout planning is assigning relative locations to a set of facilities. The best layout plan is often the one that results in the highest overall effectiveness of transactions between the facilities. Cost considerations can be a major factor in choosing a given alternative layout plan for implementation. The problem of facility layout is well covered in the literature. Some models are based on the analysis of the relative location of facilities. A graphical approach, a quantitative procedure and a quadratic assignment algorithm were developed by Buffa[1], Wimmert[2] and Hillier & Connors[3], respectively. A large mixed-integer goal-programming model for assigning offices to faculty and staff within an academic facility was formulated by Ritzman e t al.[4]. Several computer programs were developed for planning facilities layout. The Computerized Relative Allocation of Facilities Technique (CRAFT) program[5-7] calculates department centers and an estimate of total interaction cost for the input layout. The results can be used for reallocation of facilities. The limitations of CRAFT are discussed by Vollmann & Buffa[8]. Hicks & Cowan[7] modified the CRAFT program to include relocation costs as well as interaction costs in the computer program for layouts. Vollmann e t al.[9] applied a variant of the CRAFT program to the office layout of an oil company. The criterion to be minimized was the aggregate of the product of cost-weighted trips by employees by the distance walked. Volgyesi[10] used the CRAFT program as an input to the architectural design of an office. He produced a relative priority for locating departments near one another in terms of a ranking scale. Two computer programs were developed based on a rating system used to approximate the relative need for pairs of departments to be adjacent or close to each other. Computerized Relationship Layout Planning (CORELAP) was developed by Lee & Moore[l 1], and Automated Layout Design Program (ALDEP) was developed by Seehof & Evans[12]. A computer language and program for considering multiple objectives were developed by Eastman[ 13]. His approach is most appropriate at the detailed level of planning when the exact placement of desks, furniture and partitions is considered. A computerized method for allocating facilities to locations within a multifloor building was developed by Johnson[14]. The method, called " S P A C E C R A F T , " is based on the single-floor CRAFT heuristic and can incorporate the nonlinear travel times that result from having multiple floor levels. Computerized Facilities Design (COFAD) (see Tompkins & Moore[15]) is a modification
141

142

REUVLNR. [.EVARY SYLVIA and KALCHIK

of C R A F T to allow inclusion of move costs for materials-handling equipment alternatives. Evaluation studies of facilities layout techniques were done by Denholm & Brooks[16], Nugent et a/.[17], Zoller & Adendorff[18] and Tompkins & Moore[15]. Optimization and computer algorithms will sometimes produce odd department shapes. Plant Layout Analysis and Evaluation Technique (PLANET)[15], for example, often generates layouts with unusual shapes. For this reason, space-planning methodologies that are based on the visualization of the layout by the planner can be more practical and are more likely to be accepted by practitioners. In fact, an experiment in which computer models were placed in competition with individuals making visual relative locations arrangements of departments was conducted by Scriabin & Vergin[19]. The conclusion was that the visual-based methods were superior, or at least equal, to the computer-based methods. However, Trybus & Hopkins[20] found the computer-based methods to be superior. A set of computer programs called "LAYO U T " was developed by Scriabin & Vergin[ 19] to allow the individuals who participate in their experiment to interact in a conversational mode with the computer. LAYOUT performed the calculations required to evaluate the cost of any plant layout submitted by a participant in the experiment. Lawerence[21] developed a three-dimensional computer simulation model for proposed home designs. The model is used by architects to obtain a feedback from the end-users in order to be able to include their values and ideas in modifying the proposed home design. Some characteristics of 15 computer-assisted and 2 manual solution procedures to the facilities layout problem are summarized in this study. The procedures are the following: Modular Allocation Techniques (MAT)J22], CRAFT[6, 8, 10, 16, 17, 19, 23, 24], CRAFT-IV[7], CRAFT-M[7], SPACECRAFT[14], One-Dimensional Space Allocation Problem (ODSAP)[25], ALDEP[12, 16, 24, 26], CORELAP[ll, 16, 24, 27], RMP Comp. I (Richard Muther and Associates)[24], Layout Simulation Program (LSP)[18], H-63 (Hillier 1963) [17, 19], HC-63-66 (Hillier-Conners 1966)[17, 19], Graph Theory[2832], Cross Charting[33], Goal Programming[4], COFAD[15] and PLANET[15]. The following characteristics, for each of the solution procedures examined, are the base of the summary: input required, general characteristics, limitations and type of output obtained. These characteristics are described in the next section. Goal programming has been included in this survey to illustrate an approach to space planning based on multiple objectives. Multiple objectives can also be treated by a computer language and program developed by Eastman[13]. His method is most appropriate at the detailed level of planning, e.g. planning the location of furniture and partitions in an office. SUMMARY OF THE SOLUTION PROCEDURE'S CHARACTERISTICS Input required Data related to the relationship between departments are important input for planning the facilities layout of organizations with multiple departments. Departments in a bank, for example, can include personnel, credit services and loan payments, and in a factory they can include machine centers and storage areas. In most instances, these data are summarized in a Relationship Chart, commonly known as a REL Chart[l 1], or a Move Desirability Table (MDT)[17]. Certain solution procedures require information about the size of each department and the departmental move cost. Departmental move cost is occasionally described as the materials-handling cost portion of the overall cost of the facilities layout and location[6, 7, 17]. Subjective or "political" input can be handled by some of the procedures. This input is generally handled by using a fixed location of departments and some fixed building features (see Vollmann et a/.[9]. The "fixed" input can be changed at different runs of the solution procedure. Other inputs listed are unique to each procedure indicated. In general, the more input required, the better the chance of arriving at the optimal or near optimal solution. The input required by 15 of the solution procedures discussed are summarized in Table 1. PLANET has the same basic input requirements as CRAFT. The materials flow data may be inputted as either an extended parts list, a from-to chart or a penalty chart (see Tompkins &

Table

1. S u m m a r y

of input required

to the solution procedures

im'trr REqotR~
X

1.
I I

Department Identificat i o n number x2
X2 X X X X2 X2 X X X

2.

Interdepartmental flow data: REL C h a r t o r I ~ T

X2

I
X2

3.
X X3

Known d i s t a n c e / t r a v e l t i m e b a t v e e n any two locatlons or departments
X X X X X

4.

Fixed departments/ buildln8 features
X

5.

Number o f f l o o r s the buL1dlng
X

in

6.

Number o f d i f f e r e n t floor types - defined
X

O >

7.

Number o f m o d u l e s i n layout

8.

Size o f modules used as b u i l d i n g blocks X X X X X

9.

Departmental COSTS

relocation

10.

Initial

solution

11. X

Building vldth

& length

X

X X X

X X

12.

S i z e o f each d e p a r t m e n t (square feet)

Xi - O p t i o n a l required, approprlately welghted

X2 - I n c l u d e s

number o f t r i p s

X3 - T o t a l v e i B b t e d d l s t e n c e

144

RVUVI. N R. LEVARY a n d SYLVIA KALCttlK

Moore[15]). This flexibility is the only advantage of PLANET. Since COFAD is a modification of CRAFT, it has the same basic input requirements as CRAFT with the following additions: (1) an input specifying if" the alternatives are to be evaluated based on straight-line or rectilinear distances; (2) the percentage change in the from-to charts for the evaluation of solution sensitivity.

General characteristics Each solution procedure is classified according to the features most commonly associated with suboptimal facility layout procedures. There are two distinct types of suboptimal procedures: construction and improvement[19]. Construction techniques produce an assignment based on the volume of trips between all pairs of departments per unit of time and the distance between all pairs of departments per unit of time. Improvement procedures require an initial solution in addition to the input needed by the construction techniques. The methods by which multifloor layouts are handled vary according to the procedure used. The capacity of a given procedure is indicated by the maximum number of departments that it can handle. The complexity of the procedure will increase as the number of floors and/or departments increases, particularly with the use of Graph Theory[29-32] and Cross Charting[33] techniques. The general characteristics of the procedures in question are summarized in Table 2. COFAD[15] can incorporate a materials-handling system in the design of a layout. The uniqueness of PLANET[15] is in the flexibility of some of the input, as mentioned earlier. Limitations The cost of implementing a solution procedure includes the cost of computer time in an automated procedure or the cost of personnel time in a manual one. These may be very high. Occasionally, the cost of implementing a given solution procedure may be more than one could expect to save by improving the layout. It is also possible that the reduction in total distance traveled may not be large enough to justify the relocation costs[7]. Another problem is the impracticality of the solution procedure due to the complexity of implementation[29, 30, 32, 33]. Other limitations of the solution procedures discussed are the exclusion of subjective or "political" data and the generation of irregular or elongated department shapes [11, 14, 22, 24, 26]. The SPACECRAFT algorithm[14] has the limitation that departments fixed in location could prevent some possible exchanges of departments. It also can generate a layout, for a multistory building, which will result in long waiting lines near the elevators. The limitations of the solution procedures are summarized in Table 3. The limitations of COFAD[15] are identical to those of CRAFT. PLANET[15] conceals the direction of flow among departments and therefore may result in odd layouts. For this reason, this algorithm should be utilized primarily as a generator of initial layouts. Type of output obtained A block diagram layout is the most prevalent form of output. The block diagram indicates broad areas of space within the given building dimensions for suggested occupancy by various departments. The shapes or blocks obtained may be somewhat irregular, thus requiring manual adjustment. Some solution procedures produce a loose space relationship diagram that resembles a Program Evaluation and Review Technique Chart. In these cases, manual refinement is mandatory, to adjust for the physical size of the department and the physical constraints and shape of the building. Reports of varying quantity and degree of information are available from a number of the automated solution procedures. The details, which are available, range from movement time between pairs of floors, departments or building modules to the total cost per time period of movement when originating at a given department. The users of most automated solution procedures can select the degree of details required from the computer reports. The trade-off between the benefit of obtaining a given degree of details of a layout and the cost of the computer time is an important factor in determining the actual details of the report. The types of output obtained from the solution procedures are summarized in Table 4. COFAD[15] and PLANET[15] both have the same basic type

Table 2. Summary of some general characteristics of the solution procedures

GENERAL C H A R A C T E R I S T I C S

I. X X

Construction

Technique

2.

Improvement

Technique

X x x

3. X 40
40 X2 20 X

Accommodates m u l t i floor layouts X1 63 45

E

4.

Capacity

5.

Non-adjacent possible

moves f~

X I - Accomlodates

up to three

floors

X 2 - Unlimited

Table 3. Summary of some limitations of the solution procedures
CRAFT-IV
I

LIMITATIONS

I.

I m p r a c t i c a l i n terms l of c o s t , computer time, and c o m p l e x i t y of implementation

2.

Department d i s t a n c e s measured between centrotds

I
X X
X X X X

3,

Always travelling parallel to edge o f department, never diagonally

4.

Irregular or e l o n g a t e d department shapes generated

5.

Fixed d e p a r t m e n t s block exchanges

can

6.

Waiting llne may be caused by m u l t l - s t o r y

capability

7.

Reduction in lntradepartmental move costs may n o t compensate f o r r e l o c a t i o n cost

8.

Does not score or X X

numerically evaluate

layouts

9.

Strongly influenced the original layout

by

IO,

Ignores special factors or subjective/ elements of the organization

I.

Limited to exchange of adjacent departments X XI

12.

Limited t o two department changes

X 1 - Exchange 2 or 3 departments,

can be between

if d e p a r t m e n t s

are adjacent

Table 4. Summary of the type of output obtained from the solution procedures

OUTPUT OBTAINED

I.

Block Diagram

2. X X

Interdepartmental flow costs/total distance travelled Xl

3.

Space R e l a t i o n s h i p

Xl O

Diagram

4.

Special layout

5. X X2

Reports showing evaluation a t a t u 8 for current layout and each of the runs

X I

-

Requires e x t e n s i v e manual r e f i n i n g
- Physical restrictions,

for:

- p h y s i c a l 81ze of d e p a r t m e n t llmltatlons, or interferences within the area or building

X 2 - 14 reports available - can be printed between interactions

148

R~.:UVEN R. LEVARy and SYLVIA KAt.cr~Ir<

of output as CRAFT. COFAD can print a listing describing the improvements made at each iteration.
SUMMARY

This paper surveys the main characteristics of the most-used solution procedures

for facility layout problems. The tabulated characteristics can be used by practitioners for determining which solution procedure can be applied to a particular problem. The final choice of a solution procedure depends on additional factors such as budget, availability of computer hardware and software and experience of the practitioner with the solution procedure in question. Researchers can use the tabulated characteristics for determining the specifications of a new solution procedure that will avoid some of the limitations of the existing solution procedures but will have some of their advantages.
REFERENCES 1. E. S. Buffa, Sequence analysis for functional layouts. J. Indus. Engng. 6, (1955). 2. R, J. Wimmert, A mathematical method of equipment location. J. Indus. Engng. 9, (1958). 3. F. S. Hillier & M. M. Connors, Quadratic assignment problem algorithms and the location of indivisible facilities. Management Sci. 13, 42-57 (1966). 4. L. Ritzman, J. Bradford & R. Jacobs, A multiple objective approach to space planning for academic facilities. Management Sci. 25, 895-906 (1979). 5. G, C. Armour & E. S. Buffa, A neuristic algorithm and simulation approach to relative location of facilities. Management Sci. 9, 294-309 (1963). 6. E. S. Buffa, G. C. Armour & T. E. Vollmann, Allocating facilities with CRAFT. Harvard Business Rev. 42, 136-158 (1964). 7. P. E. Hicks & T. E. Cowan, CRAFT-M for layout management. Indus. Engng. 8, 30-35 (1976). 8. T. E. Vollmann & E. S. Buffa, The facilities layout problem in perspective. Management Sci. 12, 450468 (1966). 9. T. E. Vollmann, C. E. Nugent & R. L. Zarler, A computerized model for office layout. J. lndust. Engng. 19, 321-327 (1968). 10. A, S. Volgyesi, Space-age approach to space allocation. Comput. Dec'. 2, 32-35 (1970). I 1. R. C. Lee & J. M. Moore, CORELAP--Computerized Relationship Layout Planning. J. Indus. Engng. 18, 195-200 (1967). 12. J. M. Seehof & W. O. Evans, Automated layout design program. J. Indus. Engng. 18, 699-695 (1967). 13. C. M. Eastman, Automated space planning. Artificial Intelligence 4, 41-64 (1973). 14. R. V. Johnson, SPACECRAFT for multi-floor layout planning. Management Sei. 28, 407-417 (1982). 15. J. A. Tompkins & J. M. Moore, Computer Aided Layout: A User's Guide. Facilities Planning and Design Division. American Institute of Industrial Engineers (1978). 16. D. H. Denholm & G. H. Brooks, A comparison of three computer assisted plant layout techniques. Proceedings of the American Institute of Industrial Engineers Twenty-first Annual Conference and Convention, Cleveland, pp. 77-84 (1970). 17. C. E. Nugent, T. E. Vollmann & J. Ruml, An experimental comparison of techniques for the assignment of facilities to locations. Operations Res., 16, 150-173 (1968). 18. K. Zoller & K. Andendorff, Layout planning by computer simulation. A l l E Trans. 4, 116-125 (1972). 19. M. Scriabin & R. C. Vergin, Comparison of computer algorithms and visual based methods for plant layout. Management Sci. 22, 172-181 (1975). 20. T. W. Trybus & L. D. Hopkins, Human vs. computer algorithms for the plant layout problem. Management Sci. 26, 570-574 (1980). 21. R.J. Lawerence, The simulation of domestic space: users and architects participating in the architectural design process. Simulation Games 11,279-300 (1980). 22. H. K. Edwards, B. E. Gillett & M. E. Hale, Modular allocation techniques (MAT). Management S~'i. 17, 161-169 (1970). 23. T. A. Block, A note on "Comparison of computer algorithms and visual based methods tor plant layout" by M. Scriabin and R. C. Vergin. Management Sci. 24, 235-237 (1977). 24. R. D. Muther & K. McPherson, Four approaches to computerized layout planning. Indus. Engng. 2, "~9-42 (1970). 25. J. C. Picard & M. Queyranne, On the one-dimensional space allocation problem. Ops R e s 29, 371-391 (1981). 26. J. M. Seehof, W. O. Evans, J. W. Friederichs & J. J. Quigley, Automated facilities layout programs. Proceedings of the ACM National Meeting, pp. 191-199 (1966). 27. J. M. Moore, Computer program evaluates plant layout alternatives. Indus. Engng. 3, 19-25 (1971). 28. F. E. Block, On the complexity of facilities layout problem. Management Sci. 25, 280-285 (1979). 29. A. S. Carrie, J. M. Moore, R. Marke & J. J. Seppanen, Graph theory and computer aided facilities design. OMEGA 6, 353-361 (1978). 30. L. R. Foulds & D. F. Robinson, Graph theoretic heuristics for the plant layout problem. Int. J. Production Res. 16, 27-37 (1978). 31. P. H. Levin, Use of graphs to decide the optimum layout of buildings. Architect's J. Information Library 809-815 (1964). 32. J. J. Sepp~nen & J. M. Moore, Facilities planning and graph theory. Management S~'i. 17, B242-25 ~, (1970). 33. M. Schneider, Cross charting technique as a basic for plant layout. J. Indus. Engn~. I 1, 47~-483 119601.