Page 1
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
1
Rule Based Layout Planning and Its Multiple Objectives
Mohammad Komaki, Shaya Sheikh , Behnam Malakooti
Case Western Reserve University
Systems Engineering
Email: [email protected]
Abstract In this paper, we present a novel constructive approach for solving facility layout problems
considering objectives of total flow and qualitative closeness. This approach, called Head-Tail,
utilizes a set of rules to solve single and multiple row layout problems. We apply pair-wise exchange
method to improve the partially constructed layouts. The computational efficiency of Head-Tail
approach is examined by comparing it with Systematic Layout Planning. The results show that
Head-Tail method can find efficient solution with promising quality. A bi-criteria numerical
example that minimizes total flow and total qualitative closeness is also presented.
Key words: Rule-based layout planning, multi-objective optimization, Head-Tail Method
1. Introduction Effective layout designs are necessary in many different businesses such as manufacturing
facilities, and service industries. Diminishing physical space availability, high cost of re-layouts,
safety issues, and substantial long-term investments are good reasons for getting the design right
the first time. Different criteria can be considered in facility layout design including operational
costs, installation costs, utilization of space and equipment, work-force management, materials
handling, inventory work in process, inventory of incoming materials and inventory of outgoing
final products, ease of future expansion, and safety factors. Most of these objectives are difficult to
quantify. In this paper, new methods are introduced for solving multiple objective layout problems.
Researchers have used many criteria to come up with the optimal layout design. Raman et al. [1]
developed three factors of flexibility, productive area utilization and closeness gap to measure the
layout effectiveness. Rosenblatt [2] discussed a facility layout problem with two criteria of
minimizing costs and maximizing closeness rating. Fortenberry and Cox [3] discussed a multi-
criteria facility layout problem that combines quantitative with qualitative criteria and minimizes
total weighted work-flow. Malakooti [4,5,6] presented heuristic methods for solving multi-
objective facility problems.. Malakooti and Tsurushima [7] proposed an expert system for solving
multi-crtieria manufacturing problems. Islier [8] developed a genetic algorithm for solving multi
objective facility layout by minimizing transportation load, maximizing compactness of
departments and minimizing the difference between requested and available area. Pelinescu and
Wang [9] presented a hierarchical multi criteria optimal design of fixture layout. Yang and Kuo
[10] proposed a hierarchical AHP and data envelopment analysis approach for solving layout
design problem. Chen and Sha [11,12] developed a five-phase heuristic method using paired
comparison matrix for solving multi-objective facility layout problem. Sahin and Ramazan [13] and
Singh & Singh [14] have used heuristic method for solving multi-criteria layout problems. In this
paper, we develop a new rule based constructive approach called Head-Tail method. We use this
method for solving bi-criteria layout optimization problems.
Page 2
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
2
2. Head-Tail Layout Head-Tail approach starts with the pair of departments that have the highest flow as the initial
partial layout. More departments are progressively added to the existing partial layout until all
departments are assigned. The problem is to find the order of departments that minimizes the total
flow as measured in Equation (1). The selection of the location of each department is based on
qualitative or quantitative rules and other pertinent information. We call sp = (i j) as a candidate
pair of departments and sq = ( …) as a partially constructed sequence.
Generic Steps and Rules of Head-Tail Method
1. Consider the ranking of all pairs (i, j). Set the highest ranked pair as sq.
2. Consider the highest ranked unassigned pair, sp = (i , j), such that a. Both i and j are not in sq
b. Either i or j is in sq.
3. Use relevant Head-Tail Rules to combine sp and sq.
4. Repeat steps 2 and 3 until all stations are assigned to sq.
5. Try to improve upon the obtained solution by use of the Pair-Wise Exchange method.
Temporarily Discarding Unassigned Pairs: If a pair sp = (i,j) does not have any common
element with sq = (…...…), then sp = (i,j) is discarded temporarily only in the current iteration, but it
is reconsidered in the next iteration.
Permanently Discarding Assigned Pairs: If a pair sp = (i, j) has two common elements with sq =
(… i.., j…), then sp = (i,j) is discarded permanently (because both i and j are already assigned).
Consider a distance matrix, D, consisting of elements dij where dij is the distance between pairs of
departments i and j. Also consider a flow matrix, W, where each element (wij) represents the flow
from department i to department j. For every pair of departments (i, j) where i<j, create a new wij
equal to (wij + wji). These values are used in calculating the total flow. For a problem with n
departments, there will be n locations. The distances can be presented by a symmetric flow matrix;
that is, the distance from department i to department j, dij, is equal to the distance from department j
to location i, dji. The total flow for a given sequence solution is:
TF =n n
ij iji=1 j=1
w d (1)
Rules for Single Row Layout Order all pairs (i, j) in descending order of wij in a flow Matrix. Label the ordered vectors of wij by
wo. Consider sp = (i j), sq = ( …), and wo = Vector of wij in descending order of wij.
Conflict Resolution Rule: If there is a common department in sequences sp and sq, i.e. sp = (ij)
and sq = (…n i m…), then place the new department, j, next to the common department, i, in
sq on the side that has a higher ranking to its adjacent department, i.e. compare jn and jm flows
and insert j on the side that has higher flow.
Linear Mirror Property: The mirror image of sequence sq, i.e. its reverse sequence, is equivalent
to sq. Both sequences will have the same total flow. In the following, we show how to solve a
single-row layout using Head-Tail.
Example 1: Single-Row Layout Consider the information given below. Suppose the layout should be in the form of a 1x6
rectangle.
Page 3
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
3
Distance D
1 2 3 4 5 6
Locations
1 - 1 2 3 4 5
1 2 3 4 5 6
2
- 1 2 3 4
3
- 1 2 3
4
- 1 2
5
- 1
6 -
Flows V
Ranking of Flows V
A B C D E F
A B C D E F
A - 10 20 30 10 100
A - 11 8 7 11 1
B
- 50 60 10 20
B
- 5 3 11 8
C
- 10 20 80
C
- 11 8 2
D
- 50 60
D
- 5 3
E
- 10
E
- 11
F -
F -
Figure 1: Initial Data for Example 1
1. sp = (A↔F); then set sq = (A↔F)
2. sp = (C↔F); The mirror image is sp = (F↔C). Therefore, sq = (A↔F↔C).
3. Discard (B↔D) temporarily as it does not have a common element with sq.
4. sp = (D↔F); there is conflict for location of D, use the Conflict Resolution Rule to find the
location for D. This is shown by the following graphical representation. A F C
? ? D
Since the flow of (D,A) = 30 is more than the flow of (D,C) = 10, locate D between A and
F; therefore, sq = (A↔D↔F↔C). Reconsider (B↔D), the location of B should be decided. (A D F C)
? ? B
The flow of (B, F) = 20 is more than the flow of (B,A) = 10; locate B on the right side of D.
Therefore, sq = (A↔D↔B↔F↔C).
5. Consider (B↔C), discard it because it is a duplicate of elements in the sequence.
6. Consider (D↔E), the location of E should be determined. (A D B F C)
? ? E
The flow of (E,A) = 10 is equal to the flow of (E,B) = 10, therefore E can be put on either side
of D; arbitrarily, locate it on the right side of D. Therefore, sq = (A↔D↔E↔B↔F↔C).
The final layout is (A↔D↔E↔B↔F↔C). Using the above given distances and Equation 1, the
total flow of this layout is Total Flow = (10*3) + (20*5) + (30*1) + (10*2) + (100*4) + (50*2) + (60*2)
+ (10*1) + (20*1) + (10*4) + (20*3) + (80*1) + (50*1) + (60*3) + (10*2) = 1260
Rules for Multi-Row Layout We can extend the single row Head-Tail method to solve multiple row layout problems. The single
row approach expands horizontally. The multi-row layout approach uses all rules of the single row
Page 4
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
4
approach, but allows both horizontal and vertical expansions.
The following rules show how to progressively augment sq by combining existing sq = (… i
…), the partially built layout, with a new sp = (ij); i.e. find a location for department j. By use
of the following rules, all departments will be assigned and a complete solution will be obtained.
Expanding Phase Rule 1 (Attaching Sequences, Horizontal and Vertical) a) Horizontal: If there is one common department in the pair of sequences sp & sq, such that the
common department is either at the head or the tail of sequence sp and sq, then form a new sequence
by attaching sp to sq by their common department.
For example, given sq = (CBE) and sp = (AC), new sequence is sq = (ACBE).
b) Vertical: A sequence can be attached in the middle of an existing sequence at a 90° angle if the
new sequence has one common department with the existing sequence.
For example, adding the sequence (BF) to the sequence (ABCD) in Figure 2a results in
the sequence in Figure 2b.
Rule 2 (Attaching Sequences, Diagonal): If there is a common department in sequences sp and sq,
e.g. sp = (ij) and sq = (… i …), and Rule 1 fails, i.e. department j cannot be placed next to i
in sq, then insert it at a 45° angle on the side which is available and has the highest ranking priority
with respect to its adjacent departments.
Rule 3 (Extending): Keep the existing sequence as open (extended) as possible by stretching all
connected lines as straight as possible.
That is, add the new sequence to the old sequence in the form of a straight line if possible. For
example, adding (EF) to the sequence in Figure 2b results in the sequence in Figure 2c.
Reshaping Phase Rule 4 (Rotation): To fit the partial constructed layout tree into the given layout locations, if
possible, use vertical, horizontal, or diagonal rotation of branches while satisfying closeness
preferences. In Example 2, in Figure 5c, E is rotated around D resulting in Figure 5d to allow B to
fold down to C in Figure 5e.
Rule 5 (Folding): To fit the partially constructed layout tree into the given layout locations, if
possible, bend the branches of the layout tree while satisfying closeness preferences.
For example, suppose the given layout is 2x3 where the given constructed sequence is shown in
Figure 2c. Consider folding branch B-F-E. It can be folded to the right or left. Because the ranking
of (E,C) is higher than (E,A), fold the branch to the right connecting E to C as shown in Figure 2d.
Figure 2d uses the folding rule to fit into the given layout space. To do this, bend A and D upwards
and connect them as shown in Figure 2e. This is the final layout which now fits in the 2x3 space.
A B C D
A B C
F
D
A B C
F
D
E
A B C
F
D
E
B C
F E
A D
Figure 2a: Expanding
Figure 2b: Expanding
Figure 2c: Expanding
Figure 2d:
Folding Figure 2e:
Folding
Fitting Phase Rule 6 (Conflict Resolution for Multi-Row): If the generated solution by using the reshaping
rules does not fit into the given layout space, cut low priority (ranking) departments and paste them
Page 5
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
5
into the closest empty spaces based on closeness preferences.
Suppose that Figure 3a shows the sequence obtained from Phase I where the layout space is 2x3. In
Phase II folding results in Figure 3b. The sequence given in Figure 3b cannot be folded to fit the
layout’s shape of 2x3. Suppose that the departments (A,B) have the lowest flow of all attached
departments in the layout; then A is the best choice to be cut. Apply rule 6 by cutting department A
and pasting it to the open space. The result is shown in Figure 3c. A B C
F
D E
A B C
F
D
E
A
B C
F
D
E Figure 3a:Given Solution Figure 3b:Folding Figure 3c:Cut and Paste A
Example 2: Head-Tail Multi-Row Layout Consider the information given below. Suppose the layout should be in the form of a 2x3 rectangle.
Distance D
1 2 3 4 5 6
Locations
1 - 1 2 1 2 3
1 2 3
2
- 1 2 1 2
4 5 6
3
- 3 2 1
4
- 1 2
5
- 1
6 -
Flows V
Ranking of Flows V
A B C D E F
A B C D E F
A - 10 20 30 10 100
A - 11 8 7 11 1
B
- 50 60 10 20
B
- 5 3 11 8
C
- 10 20 80
C
- 11 8 2
D
- 50 60
D
- 5 3
E
- 10
E
- 11
F -
F -
Figure 4: Initial Data for Example 2
Expanding Steps
1. sp = (AF). sq = (AF) 2.sp = (CF) or sp = (FC). sq = (AFC)
3. sp = (BD). Ignore sp temporarily. 4. sp = (DF). sp can be attached to sq, see Figure 5a.
5. sp = (BD). B can be attached to D, see Figure 5b.
6. sp = (DE). E can be attached to D. Since (E,C) has a higher flow than (E,A), E is attached to
the right side of D. See Figure 5c
All departments are assigned; reshape to fit the layout to the 2x3 block requirement.
Reshaping Steps
7. Consider Figure 5c. (E,C) has a flow of 20 compared to a flow of 10 for (E,A). (B,C) has a
flow of 50 compared to a flow of 10 for (B,A). Therefore, the advantage of having B next to C
(an improvement of 40) outweighs the advantage from having E next to C (an improvement of
10). Therefore, in Figure 5c, E can be rotated from left to the right side. This rotation will allow
B to fold towards (closer) to C instead of A. The result of this rotation rule is shown in Figure
5d.
8. In Figure 5e, B is folded down toward C.
The final solution layout is shown in Figure 5f. The total flow for this solution is Total Flow =
Page 6
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
6
700.
A F C
D
A F C
D
B
A F C
D
B
E
Figure 5a: Expanding Figure 5b: Expanding Figure 5c: Expanding
A F C
D
B
E
A F C
D
B
E
A F C
D B E
Figure 5d: Rotation of E Figure 5e: Folding Figure 5f: Solution
Example 3: Comparing Head-Tail Layout and SLP Consider a facility consisting of seven departments, each with individual space requirements as
shown in the table below. Each department can be broken up into units or blocks of 100 square
feet. The total allotted area for the new facility is 2400 square feet and must be laid within a
rectangular space with a dimension of 4x6 blocks. Each department can fit into a square or a
rectangle (e.g. a department can be presented by a 1x6 or 2x3 block shape). Department 1 2 3 4 5 6 7 Sum
Square feet 600 200 400 200 500 100 400 2400
No. of Blocks 6 2 4 2 5 1 4 24
The matrix below shows the given desired proximity or closeness for each pair of departments. We
first use SLP to find a layout solution to this problem for a. 400x600 facility, and b. 200x1200
facility.
1 2 3 4 5 6 7
1 A O A I A A
2 U E O E X
3 X U U E
4 U O O
5 I I
6 X
7
Figure 6 show the initial solution where each department is presented by a node, and the desired
closeness for each pair of departments is presented by lines that connect the two departments.
Figure 7 shows the first iteration of the rearrangement of the relationship diagram to minimize the
strain on these bands.
1
3
2 7
5
6
4 Figure 6: Initial graph to present the problem Figure 7: First iteration to improve the total closeness
Figure 8 shows the next iteration. Re-arrangement of the network should be repeated until no more
34
7
6 5
2
13
4
7
6 5
2
13
4
7
6 5
2
13
4
7
6 5
2
1
Page 7
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
7
improvement can be achieved. Once the final graphical relationship is achieved, the space
requirements for each department are considered. Figure 9 shows the area of each department.
2
5
4
6
3
7
1
1
1
1 1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
6
6
6
2
2
2
2
2
2 3
3
3 3
3
3
3
3
3
3
3
3
4
4
4 4
4
4
7
7
7
7
7
7
7
7
7
7
7
5
5 5
5
5
5
5 5
5
5
5
Figure 8: Further rearrangement to improve the total
closeness
Figure 9: Relationship diagram showing departmental
space requirements
Now we can collapse and rotate the spatial relationship diagram shown in Figure 9 to fit into the
4x6 space. This is shown in Figure 10. For the 2x12 block facility, the solution is shown in Figure
11.
2 2 1 1 4 4
5 6 1 1 7 7 5 5 5 4 2 1 1 1 7 7 3 3
5 5 1 1 7 7 5 5 6 4 2 1 1 1 7 7 3 3
5 5 3 3 3 3
Figure 10: Relationship diagram for a
4x6 area
Figure 11: Layout solution for a 2x12 area
The above solution generated by SLP is based on space requirements, availability, closeness
preferences, and other relative factors. The A, E, I, O, U, and X ratings can be converted to the
numerical values, 5, 4, 3, 2, 1, and 0, respectively. Then the ranking of the pairs of departments
will be performed. Note that pairs that tie have same ranking.
Rankings of Pairs (i,j)
1 2 3 4 5 6 7
1 - 1 11 1 8 1 1
2 - 15 5 11 5 X
3 - X 15 15 5
4 - 15 11 11
5 - 8 8
6 - X
7 -
The construction of the layout by the Head-Tail method is as follows.
Expanding Steps
Step 1: sp = (12). sq = (12) Step 2: sp = (14). sq = (412).
Steps 3, 4, and 5: Departments 6, 7, and 3 are added; see Figure 12.
Step 6: Next consider sp = (15). Department 5 can be located in four possible locations
diagonally from department 1 as shown in Figure 13, Step 6. Because pairs (5,2) and (5,6) have the
highest ranking, it will be located next to 2 and 6.
6 1
2
4
6 1
2
4
7
6 1
2
4
7 3
6 1
2
4
7 3
5
Page 8
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
8
Step 3: Connect 6 to 1 Step 4: Connect 7 to 1 Step 5: Connect 3 to 7 Step 6:
Connect 5 to 6 and 2
Figure 12: Graphical Representation of layout tree Steps 3, 4, 5, and 6.
Since all departments are assigned, the expansion phase is complete.
Reshaping Steps
Step 7: Consider the next pair, (4↔6). Departments 4 and 6 cannot be folded. Consider the next
pair, (4↔7). Departments 4 and 7 cannot be folded. Consider the next pair, (2↔3). Fold 3 closer
to 2 (see Step 7 in Figure 13).
Fitting Steps
Step 8: Present the actual size and shape of the departments.
Step 9: Move the departments to fit in the layout size of 4x6 units.
6 1
2
4
7
3 5
5
5
5
6
5
5
1
1
1
1
1
1
2 2
4 4
3
3
3
3
7
7
7
7
5 5 2 2 7 3
5 6 1 1 7 3
5 4 1 1 7 3
5 4 1 1 7 3
Step 7: Connect 3 to 2 Step 8: Insert Spaces Step 9: Final Layout
Figure 13: Graphical Representation of Steps 7, 8, and 9
The final layout is shown in Step 9 of Figure 13. The center of each department is presented by a
“●”. The distances between each pair of departments are shown in the following table.
1 2 3 4 5 6 7
1 200 300 200 310 250 200
2 400 400 350 250 300
3 500 510 450 100
4 210 150 400
5 80 410
6 350
7
The total preferential closeness is: Total Closeness = n n
i=1 j=1
Closeness(i,j)*Distance(i,j) = 13,620
Comparison of Head-Tail and SLP Solutions: The above example was solved by SLP method.
The layout and the distances for the SLP solution are shown in Figure 14. The total closeness for
the SLP solution is 15,500. The Rule-based Planning solution (Example 3) was 13,620 which is
considerably better than the solution found by the SLP method.
1 2 3 4 5 6 7
1 300 300 300 330 150 250
2 600 400 230 150 550
3 400 390 450 250
4 630 450 150
5 180 480
6 400
7
Page 9
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
9
2 2 1 1 4 4
5 6 1 1 7 7
5 5 1 1 7 7
5 5 3 3 3 3
Figure 14: SLP Layout Solution
3. BI-CRITERIA LAYOUT PLANNING In bi-criteria facility layout problem, the two objectives are:
Minimize Total Flow f1 = n n
ij iji=1 j=1
w d
Minimize Total Closeness f2 = n n
ij iji=1 j=1
w p
For example, suppose that the flow between departments A and B is very high and from the total
flow point of view should be located very close to each other. However, from a safety point of
view, they should be very far from each other. The above bi-criteria problem can be used to handle
such conflicting objectives.
Suppose that an additive utility function can be used to rank multi-criteria layout alternatives. Let
w1 and w2 be the weights of importance of the two objective functions, respectively. First find the
normalized values of the two objective functions. To normalize the flow information, find the
largest flow (dij,max) and smallest flow (dij,min), and then use the formula shown below to normalize
all flow values presented by dij: they will be between 0 and 1. Similarly, closeness information can
be normalized between 0 and 1 using the following formula where pij,max is the largest and pij,min is
the smallest pij values.
vij’ = ij ij,min
ij,max ij,min
(w - w )
(w - w ) pij’ =
ij ij,min
ij,max ij,min
(p -p )
(p -p )
Then for each pair of departments (i,j), calculate
zij = w1dij’ + w2pij’
The matrix of zij values is called the composite matrix, Z. The additive utility function (U) is
defined as the weighted sum of the flow and the closeness values using weights of importance of
w1 and w2 for the flow and closeness, respectively. In order to minimize the additive utility
function U = w1f1’ + w2f2
’,
Minimize U = n n
ij 1 ij 2 iji-1 j=1
d (w d ' + w p ') =n n
ij iji=1 j=1
d z (2)
The value of zij = w1dij’ + w2pij’ is constant for i, j, and given weights. Therefore, they only need to
be calculated once for each pair of departments. Therefore, for the generated Z matrix, the bi-
criteria layout problem can be solved by Head-Tail method. The alternative with the smallest U
value is the best layout.
Example 4: Bi-Criteria Layout
Distances for the SLP Solution
Page 10
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
10
For a four department facility layout problem presented in Figure 14, find the best bi-criteria layout
solution. Suppose that the decision maker’s weights of importance for flow and closeness
objectives are w1 = 0.3 and w2 = 0.7, respectively. Consider using 0 to 5 numerical ratings for X to
A qualitative ratings. Location Flow Matrix
1 2 3 4 A B C D
A - 100 40 10
Assignment B - 60 30
A B C D C - 90
D -
Qualitative Matrix Converted Qualitative Matrix
A B C D A B C D
A - X/0 E/4 A/5 A - 0 4 5
B - U/1 E/4 B - 1 4
C - X/0 C - 0
D - D -
Figure 15: Initial data for the bi-criteria layout problem
First convert qualitative closeness data to quantitative data as presented in Figure 15. Then
normalize flow matrix and closeness matrix values as shown in Figure 16. Normalized Flow
Matrix
Normalized Qualitative
Matrix
(vij’) = (vij-10)/(100-10) (pij’) = (pij-0)/(5-0)
A B C D A B C D A B C D
A - 1 0.333 0 A - 0 0.8 1 A - 0.3 0.66 0.7
B - 0.556 0.222 B - 0.2 0.8 B - 0.307 0.627
C - 0.889 C - 0 C - 0.267
D - D - D -
Figure 16: Normalized flow and qualitative closeness values
A composite matrix, Z, can be generated by using zij = w1vij’ + w2pij’= 0.3vij’ + 0.7pij’.
Solving the Bi-Criteria Problem by the HEAD-TAIL Method: First rank all pairs in descending
order of zij values: they are (A,D), (A,C), (B,D), (B,C), (A,B), and (C,D).
The steps of HEAD-TAIL for assigning departments are:
1) (AD); 2) (CAD); 3) (CADB)
This is the final solution, and its mirror is (BDAC). The objective values for this layout are:
f1 = 2*100 + 1*40 + 1*10 + 3*60 + 1*30 + 2*90 = 640 and f2 =2*0+1*4+1*5+3*1+1*4+2*0 =16
No improved solution can be found by applying the Pairwise Exchange Method to the final solution
of this example.
Using the additive utility as the performance measure for each facility, the first iteration of the Pair-
Wise Exchange method produces the results shown below.
Coefficient Facility
Pairs
Initial
distance
Pair-Wise Exchange new distance
AB AC AD BC BD CD
0.3 AB 1 1 1 2 2 3 1
0.66 AC 2 1 2 1 1 2 3
0.7 AD 3 2 1 3 3 1 2
Page 11
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
11
0.307 BC 1 2 1 1 1 1 2
0.627 BD 2 3 2 1 1 2 1
0.267 CD 1 1 3 2 2 1 1
U = ij iji j
zd 5.548 1722 4.682 4.828 4.828 4.748 1788
The lowest score is 4.682 for swapping A and C.
Improved Layout: C B A D
The swap is made and the initial layout for the second iteration is as follows:
Coefficient Facility
pairs
Initial
distance
Pair-Wise Exchange new distance
AB AC AD BC BD CD
0.3 AB 1 1 1 2 2 1 1
0.66 AC 2 1 2 3 1 2 1
0.7 AD 1 2 3 1 1 1 2
0.307 BC 1 2 1 1 1 3 2
0.627 BD 2 1 2 1 3 2 1
0.267 CD 3 3 1 2 2 1 3
U= ij iji j
zd 4.682 4.402 5.548 4.748 4.682 4.762 4.402
The lowest score is 4.402 for swapping A and B. The new layout is.
Improved Layout: C A B D
Next iteration is:
Coefficient Facility
pairs
Initial
distance
Pair-Wise Exchange new distance
AB AC AD BC BD CD
0.3 AB 1 1 2 1 1 2 1
0.66 AC 1 2 1 3 1 1 2
0.7 AD 2 1 3 2 2 1 1
0.307 BC 2 1 1 2 2 3 1
0.627 BD 1 2 1 1 3 1 2
0.267 CD 3 3 2 1 1 2 3
U= ij iji j
zd 4.402 4.682 4.828 1788 1722 4.042 4.682
The lowest score is 4.042 for swapping B and D. The new layout is shown below.
Improved Layout: C A D B
Continuing with the next iteration:
Coefficient Facility
pairs
Initial
distance
Pair-Wise Exchange new distance
AB AC AD BC BD CD
0.3 AB 2 2 3 1 1 1 2
0.66 AC 1 3 1 2 2 1 1
0.7 AD 1 1 2 1 1 2 1
0.307 BC 3 1 2 3 3 2 1
0.627 BD 1 1 1 2 2 1 3
0.267 CD 2 2 1 1 1 3 2
U= ij iji j
zd 4.042 4.748 4.468 4.762 4.762 4.402 4.682
Page 12
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
12
No more improvement is possible, stop. The layout remains with f1 (total flow) = 640 and f2 (total
closeness) = 32. Final Layout: C A D B
Efficient Frontier By systematically varying weights (w1, w2) in Equation (2), a set of efficient bi-criteria alternatives
can be generated. To do this, first generate the Z matrix for each given set of weights and then use
head-Tail method to solve the problem. Note that for a given set of weights, when PWE method is
used, alternatives generated in different iterations of PWE method may also be efficient and should
be recorded. That is, each Pairwise Exchange solution in each iteration of PWE method can be
efficient. Consider the set of four weights given in the first row of the following table. For each
given set of weights, solve the problem by Head-Tail method. The generated solution is recorded
under each set of weights. The set of generated four alternatives are all efficient. (w1, w2) (0.99, 0.01) (0.6, 0.4) (0.5, 0.5) (0.01, 0.99)
Layout ABCD BACD DBAC CADB
Min. f1 420 460 580 640
Min. f2 32 28 20 16
Efficient? Yes Yes Yes Yes
Conclusion
In SLP, the process of improving the initial graph of the layout becomes very difficult as the
number of departments increases. Also the evaluation process in SLP is subjective and depends on
the layout designer. A different approach for solving layout problem is by use of Head-Tail Layout
method which is a powerful tool for solving both structured and ill-structured layout problems.
Head-Tail can solve very large layout problems with minimal computational efforts. Because this
method is Head-Tail, it can incorporate complicated constraints while developing the layout
solution. Furthermore, it can be used in conjunction with Expert systems approaches to solve
layout problems.
References: [1] Raman D., S. V. Nagalingam, G. C.I. Lin, “Towards measuring the effectiveness of a facilities layout”, Robotics
and Computer-Integrated Manufacturing, doi:10.1016/j.rcim.2007.06.003, 2008.
[2] Rosenblatt, M. J., "The facility layout problem: a multi-goal approach." International Journal of Production
Research, 17, 1979, pp. 323-332.
[3] Fortenberry, J. C., J. F. Cox, “Multiple criteria approach to the facilities layout problem”, International Journal of
Production Research, Vol. 23, Issue 4, 1985, pp.773-782.
[4] Malakooti, B., "Multi-Objective Facility Layout: A Heuristic Method to Generate All Efficient Alternatives",
International Journal of Production Research, Vol. 27, No. 7, 1989, pp. 1225-1238.
[5] Malakooti, B., “Computer Aided Facility Layout Selection with Applications to Multiple Criteria Manufacturing
Planning Problems.” Large Scale Systems: Theory and Applications, Special Issue on Complex Systems Issues in
Manufacturing, Vol. 12, 1987, pp. 109-123.
[6] Malakooti, B., A. Tsurushima, "An Expert System Using Priorities for Solving Multiple Criteria Facility Layout
Problems." Inter. Journal of Production Research, Vol. 27, No. 5, 1989, pp. 793-808.
[7] Yang, Z., B. Malakooti, “A Multiple Criteria Neural Network Approach for Layout of Machine -Part Group
Formation in Cellular Manufacturing.” Third IE Research Conference, 1994, pp. 249-254.
[8] Islier A. A., “A genetic algorithm approach for multiple criteria facility layout design”, International Journal of
Production Research, 1998, Vol. 36, No. 6, pp. 1549-1569.
[9] Pelinescu D. M., M. Y. Wang, "Multi-objective optimal fixture layout design", Robotics and Computer-Integrated
Manufacturing, Vol. 18, 2002, pp. 365-372.
Page 13
Proceedings of the 2014 ICAM, International Conference on Advanced and Agile Manufacturing, Held at Oakland
University, Rochester, MI 48309, USA Copyright © 2014, ISPE and ISAM USA.
13
[10] Yang T., Ch. Kuo, “A hierarchical AHP/DEA methodology for the facilities layout design problem”, European
Journal of Operational Research, 147, 2003, pp. 128–136.
[11] Chen C.-W., D. Y. Sha, “Heuristic approach for solving the multi-objective facility layout problem”, International
Journal of Production Research, Vol. 43, No. 21, 1, November 2005, pp. 4493–4507.
[12] Sha, D., C. Chen, "A New Approach to the Multiple Objective Facility Layout Problem", Integrated
Manufacturing Systems, Vol. 12, No.1, 2001, pp. 59-66.
[13] Şahin, Ramazan. “A Simulated Annealing Algorithm for Solving the Bi-objective Facility Layout Problem.”
Expert Systems with Applications, Vol. 38, Issue 4, 2011, pp. 4460-4465.
[14] Singh S. P., V. K. Singh, “Three-Level AHP-Based Heuristic Approach for a Multi-Objective Facility Layout
Problem.” International Journal of Production Research, Vol. 49, Issue 4, 2011, pp. 1105-1125.