Computer Aided Layout

The Facility Layout Problem

• Given the activity relationship as well as the space
of the department, how to construct plan the
layout of the facility

• The basis of the layout planning is the closeness

ratings or material flow intensities
– Minimize the flow times distance
– Maximize the closeness (adjacency)

• For most practical real world instances, the

computational complexity has results in various
heuristics
 Computer-Aided Layout
• Computer-aided layout supports layout design.
– Generate many layout alternatives in a short time.
– Helps conduct “what if?” and sensitivity analysis.
– Modeling the problem helps understand the system
perspective.
• Commercial software
– Most algorithms have not yet been commercialized
although they are available as “research” code and
used by consultants.
– Educational software is available.
• We’ll use one package.
 Computer-Aided Layout
Inputs
1. From-To chart and/or
2. Activity relationship chart (you may have the
flexibility to map the A,E,I,O,U,X scale to
different numerical scales).
 Can change results by changing the scaling.
 Can have negative values for relationships (e.g.,
X).
 Computer-Aided Layout
Layout development criteria.
– When constructing a layout, an algorithm implemented
on a computer needs specific criteria to compare
alternative layouts.
– These criteria may differ depending on the form of input
data characterizing flow/relationships.
– The criteria must be computable (quantitative) and is
referred to as the objective function of the “layout
problem”.
 Computer-Aided Layout
• From-to chart as input data.
– The objective will be to construct a layout to minimize
required movement costs over a specific time period.
 Computer-Aided Layout
• The objective function can be expressed as:

m m
z  f ij cij d ij
i 1 j 1

• This is called a distance-based objective function.

• The fij and cij are input data.
• If the cij= 1 for all i and j, z= total travel distance per time
period.
 Distance Measure
• The dij ‘s are normally rectilinear centroid to centroid
distance.
• The dij change as a function of department location and
shape.

• Rectilinear distance
If the centroidof department1is locatedat (x 1, y1)
andthe centroidof department2 is locatedat (x 2, y 2)

thend12 | x 1  x 2 |  | y1  y 2 |
 Computer-Aided Layout
• Activity relationship chart as input.
– The activity relationship between departments i and j must
first be quantified to get fij, where represents the relationship
strength on some numerical scale.
• One possible objective function:
m m
z    fij xij
i 1 j 1

 1 if depts.i and j are adjacent

wherex ij  
0 otherwise
 Computer-Aided Layout
• This is called an adjacency based objective.
• Since the numerical scale is arbitrary another
objective function is often used. It is called an
efficiency rating.
• It is a relative measure of a layout with respect to the
best (sometimes not possible) objective function
value.
 “Optimizing” a Layout
• Finding the optimal layout that minimizes either of the
objective functions described is a very difficult
problem (computationally).
• General methods for exploring the “space” of layouts
have been applied.
– Pairwise exchange method – A general approach used to
improve an existing layout.
– Graph-based method
– Other methods
• “Metaheuristics”
• Genetic Algorithm
 Pairwise Exchange Method
• This is a heuristic search procedure. It is not
guaranteed to find the optimal layout (except under
certain circumstances).
• In general, this is a method that should be
computerized but is applicable by hand in small
examples.
• We’ll examine an example with the cost of material
movement used as a criteria to evaluate layouts
(distance based objective).
– If all costs are the same (=1) this is total distance.
– Activity relationship charts and adjacency based objectives can
be used.
 Pairwise Exchange Method
• Algorithm
1. “Layout” = Initial layout. “Cost” = Distance based cost for
“Layout”.
2. For each possible pairwise exchange of departments in
“Layout”, calculate the total distance based cost.
3. If the lowest cost from the pairwise exchanges is > “Cost”
got to 4.
 Else “Layout” = Layout with the lowest cost.
 “Cost” = Lowest cost from step 2.
 Go to 2.
4. End with “Layout” and “Cost” as the final layout and cost.
 Pairwise Exchange Method
• Example from text.
– Assume distances are centroid to centroid.
– Assume travel from dept i to j is symmetric.

From\ 1 2 3 4
To
1 -- 10 15 20
2 -- 10 5
3 -- 5
4 --
 Pairwise Exchange Method
• Phase I: Construct Phase Initial Solution
(1,2,3,4)
• Phase II: Improvement – Pair Wise
Exchange
• a) Exchange two departments
• b) If results in better solution, accept; go to a)
• otherwise stop
Pairwise Exchange Method
Pairwise Exchange Method
Pairwise Exchange Method
Pairwise Exchange Method
 Pairwise Exchange Method
• Limitations
– No guarantee of optimality,
• The final solution depends on the initial layout
• Leads to suboptimal solution
– Does not consider size and shape of
departments
• Additional work has to be re-arrange the department
if shaper are not equal
 Graph-based Method
• Often applied when using an adjacency-based
objective.
• A method that uses a graph (nodes connected by arcs)
to represent departments and their adjacencies to
other departments.
– Departments = nodes.
– If an arc connects two departments, they are adjacent.
• Arcs have values that represent adjacency relationships.
– Finding the “best” layout amounts to finding the best graph.
Graph-based Method
 Graph-based Method
• Procedure for generating an adjacency graph.
1. Find the highest valued adjacency relationship as the
starting nodes connected by an arc (break ties randomly).
2. The third department selected will have the largest total
relationship value with the two departments selected in step
1.
3. The next departments to enter are selected one at a time
based on the highest total relationship values added when
the department is placed on the “face” of the existing graph.
 A face is a region bounded by arcs of the graph.
Graph-based Method
 Specific Computerized Layout
Algorithms
• Objective – Understand how different
quantitative concepts can be applied to
generate layout alternatives.
– Discrete and continuous departments.
– Different algorithms.
 Specific Computerized Layout
Algorithms
• CRAFT – Computerized Relative Allocation of Facilities
Technique.
– One of the oldest implementations of computerized layout.
Well known in the facilities planning literature.
• Inputs
– From-To chart.
– Cost matrix.
– Initial layout.
• Objective
– Distance based.
• Department representation.
– Discrete grids.
– No shape restrictions.
 Specific Computerized Layout
Algorithms
• CRAFT automatically implements a modified pairwise
interchange method.
– Many details must be addressed.
• CRAFT Algorithm
1. Start with an initial layout with all departments made up
of individual square grids (each representing the same
amount of space).
2. Estimate the best two-way department exchange by
assuming department centroids exchange exactly.
 Dept. i and j exchange
– New centroid i = centroid j.
– New centroid j = centroid i.
 Only consider exchanging adjacent departments.
– Why?
 Specific Computerized Layout
Algorithms
• CRAFT Algorithm
3. Execute the exchange if the estimated cost of the best
exchange in (2) is lower than the best cost found so far.
 The actual result of the exchange is problem dependent.
4. If the estimated cost of the best exchange in (2) is higher
than the best cost found so far stop.
 Else, go to 1.
CRAFT Example
 CRAFT Example
• Initial Layout
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
2 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
3 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
4 2 2 2 2 2 3 3 3 3 3 5 5 7 7 7 7 7 7
5 2 2 2 2 2 3 3 3 3 3 5 5 5 5 5 5 5 5
6 2 2 2 2 2 3 3 3 3 3 5 5 5 5 5 5 5 5
7 2 2 2 2 2 4 4 4 4 6 6 6 6 6 6 6 5 5
8 4 4 4 4 4 4 4 4 4 6 6 6 6 6 6 6 6 6
9 4 4 4 4 4 4 4 4 4 6 6 6 6 6 6 6 6 6
10 4 4 4 4 4 4 4 4 8 8 8 8 8 6 6 6 6 6

Initial layout cost = 2492 unit

 CRAFT Example
• Flow between Centroid
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
2 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
3 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
4 2 2 2 2 2 3 3 3 3 3 5 5 7 7 7 7 7 7
5 2 2 2 2 2 3 3 3 3 3 5 5 5 5 5 5 5 5
6 2 2 2 2 2 3 3 3 3 3 5 5 5 5 5 5 5 5
7 2 2 2 2 2 4 4 4 4 6 6 6 6 6 6 6 5 5
8 4 4 4 4 4 4 4 4 4 6 6 6 6 6 6 6 6 6
9 4 4 4 4 4 4 4 4 4 6 6 6 6 6 6 6 6 6
10 4 4 4 4 4 4 4 4 8 8 8 8 8 6 6 6 6 6
• Iteration 1, exchange Dept E and F
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
2 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
3 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
4 2 2 2 2 2 3 3 3 3 3 6 6 7 7 7 7 7 7
5 2 2 2 2 2 3 3 3 3 3 6 6 6 6 6 6 6 6
6 2 2 2 2 2 3 3 3 3 3 6 6 6 6 6 6 6 6
7 2 2 2 2 2 4 4 4 4 5 5 5 5 5 5 6 6 6
8 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 6 6 6
9 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 6 6 6
10 4 4 4 4 4 4 4 4 8 8 8 8 8 5 5 6 6 6

layout cost = 2393 unit (initial = 2492 unit)

• Switch department B and C
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
2 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
3 1 1 1 1 1 1 1 1 1 1 7 7 7 7 7 7 7 7
4 3 3 3 2 2 2 2 2 2 2 6 6 7 7 7 7 7 7
5 3 3 3 3 2 2 2 2 2 2 6 6 6 6 6 6 6 6
6 3 3 3 3 2 2 2 2 2 2 6 6 6 6 6 6 6 6
7 3 3 3 3 2 4 4 4 4 5 5 5 5 5 5 6 6 6
8 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 6 6 6
9 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 6 6 6
10 4 4 4 4 4 4 4 4 8 8 8 8 8 5 5 6 6 6

layout cost = 2288 unit (iteration 1 = 2393 )

 Specific Computerized Layout
Algorithms
• BLOCPLAN
• Inputs
– Activity relationship chart.
– From-To chart (if desired).
– Initial layout.
• Objective
– Adjacency based or distance based.
• Department representation.
– Continuous.
– Restricted to horizontal bands across the facility.
– Maximum number of departments = 18 (in the
software).
 Specific Computerized Layout
Algorithms
• BLOCPLAN automatically implements a
pairwise interchange method.
– Can also accommodate manual exchange.
• BLOCPLAN has a straightforward method of
converting an activity relationship chart to a
numerical rating.
– Default scale: A=10, E=5, I=2, O=1, U=0, X=-10.
– Can modify the scale if desired.
 Specific Computerized Layout
Algorithms
• BLOCPLAN also will convert a from-to chart
to an activity relationship chart.

From-To Flow Between

A B C D A B C D

A X 10 5 4 A X 11 7 9

B 1 X 1 20 B X 4 28

C 2 3 X 10 C X 20

D 5 8 10 X D X
Specific Computerized Layout
Algorithms
• BLOCPLAN also computes a “Rel-Dist” value for a
layout, which is:

m m

 f d
i 1 j 1
ij ij

– The fij are the relationship values.

 Specific Computerized Layout
Algorithms
• MULTIPLE – MULTI-floor Plant Layout
Evaluation
• Inputs
– From-To chart (if desired).
– Cost matrix.
– Initial layout.
– Lift data (if using for a multi-floor layout).
• Objective
– Distance based.
• Department representation.
– Discrete grid.
 Specific Computerized Layout
Algorithms
• MULTIPLE automatically implements a
pairwise interchange/exchange method.
– Can also implement a “metaheuristic” search
procedure called simulated annealing.
• MULTIPLE overcomes inflexibilities existing
in other methods presented.
– CRAFT – Can only exchange adjacent departments.
– BLOCPLAN – Departments restricted to “bands” across
the facility.
 Specific Computerized Layout
Algorithms
• MULTIPLE also is applicable to multi-floor facility layout.
It overcomes extensions of other methods to multi-
floor layout.
– CRAFT
• Department splitting occurs.
• No consideration of lift locations.
• Independent floor layout.
• MULTIPLE allows for more departmental shape
flexibility than BLOCPLAN and more department shape
control than CRAFT.
• MULTIPLE also can include facility constraints such as
walls, fixed department locations, and obstructions in a
straightforward manner.
 Specific Computerized Layout
Algorithms
• Fundamental basis for the layouts generated by
MULTIPLE.
– A “space filling curve” or hand generated “conforming curve”
is used.
– A space filling curve is a mathematical entity with a precise
definition.
– MULTIPLE uses one type of space filling curve called a Hilbert
curve.
– Intuitively think of a space filling curve as a path traveled when
moving from grid to grid in a layout. The path will pass through
the center of each grid and can only move to adjacent grids.
All turns in the path must be right-angle turns.
 Specific Computerized Layout
Algorithms
• Example of a space filling curve.
 Specific Computerized Layout
Algorithms
• How does MULTIPLE use these curves to
generate layouts?
– With a curve established for a facility.
1. Define a department layout sequence.
2. Place the departments in the facility in department
sequence along the curve, grid by grid.
 Specific Computerized Layout
Algorithms
• Example.
 Specific Computerized Layout
Algorithms
• Optimizing.
– MULTIPLE can implement a pairwise
exchange/interchange algorithm.
• All possible exchanges can be examined since the
exchanges are exchanges in sequence position.
• Exchanging departments of different sizes and the
impact on the rest of the layout is handled by the space
filling curve.
 Controlling Department
Shape
• MULTIPLE incorporates a method to control
department shape.
• Let
– Pi denote the length of the perimeter of department i.
– Ai denote the area of department i.
• Overall principle – For a fixed department area, as the
department shape becomes more “irregular”, its
perimeter gets larger.
 Controlling Department
Shape
• Example – MULTIPLE represents departments as grids.

is more irregular than

P= 20 sides P= 12 sides
 Controlling Department
Shape
• The minimum perimeter occurs for a department when
it is square in shape.
– Assumes the department is non-circular, and represented
as a grid made up of individual square units.
• Let Pi* = minimum perimeter occurring when
department i is square.

Pi* 4 Ai

• If a square represents the ideal department shape, a

normalized shape measure for a department can be
constructed.
 Controlling Department
Shape
• Let Ωi = Normalized shape measure for department i.
Pi Pi
i  * 
Pi 4 Ai

• Ωi ≤ 1.50 is recommended for MULTIPLE.

• A separate upper limit for Ωi for each department can be
specified.
– Exchanges resulting in Ωi ≥ upper limit are rejected.
 Multi-Floor Facility Layout
• More common in older cities and buildings (e.g.,
Detroit) and in countries where land is very expensive.
• Less flexibility of use of total facility space.
– A multi-floor building may have enough total space but may
not be able to accommodate departments on different levels
without splitting.
 Multi-Floor Facility Layout
• Additional decisions/constraints.
– The number and location of vertical material handling devices.
– Restriction of certain departments to specific floors.
• First floor restrictions due to floor loading capacity.
• Ceiling heights.
• Heat/chemical use and/or generation.
– Departments may have to be kept contiguous.
 Heuristics in Computer-Aided
Layout
• The mathematical description or formulation
of the layout problem is mathematical
specification of the “optimization” problem
that a computer-aided layout algorithm (e.g.,
CRAFT, BLOCPLAN, MULTIPLE) attempts to
solve.
 Heuristics in Computer-Aided
Layout
• Example for a simplified problem called the quadratic
assignment problem or QAP.
– Deciding what locations to assign departments (or facilities to.
All departments can be located at any site.
Let,
f ij  Materialflow betweendepartment
s i and j.
d hk Distancebetweensitesh andk.
c ij Costof movingoneunit of materialonedistanceunit between
department
s i and j.

m m m m
Minimizez      c ijf ijd hk x ih x ik
i 1 j1 h 1 k 1

Subjectto :
m

x
i 1
ih 1 for all h (onlyonedept.assigned
per site).

x
h 1
ih 1 for all i (eachdept.assigned
to a site).

x ih  {0,1}
 Heuristics in Computer-Aided
Layout
• The optimization problems that CRAFT, BLOCPLAN and
MULTIPLE attempt to solve are more complicated due
to department size considerations.
• These problems share one feature in common.
– They are very difficult to solve (find the best or a best solution
wrt to the given objective function).
• Procedures exist but they can become so computationally
expensive that all of the world’s computing power cannot solve
them.
 Heuristics in Computer-Aided
Layout
• Example – MULTIPLE
– MULTIPLE simplifies the layout problem by restricting layouts
to be constructed in sequence on the space filling curve.
– MULTIPLE searches for the best or “optimal” sequence.
– Finding this sequence can also be very difficult as the number
of departments increases.
• Example – 25 departments. How many sequences?
 There are A lot of
Possibilities
There are 25! different schedules
The first task has 25 choices, the
second has 24 choices….

25! = 15511210043330985984000000

25! Pennies would cover the whole state

Of Texas to a height of over 6,000 miles
(Hopp & Spearman1996)
Now that’s Big!