You are on page 1of 10

CELLULAR MANUFACTURING

= application of Group Technology


to convert a manufacturing system into cells

Cell = cluster of machines/processes located in close proximity


- dedicated to manufacture of a part family
- share process similarities

cellular approach:

reduces set-up time - (using part family sequencing and tooling)


reduces flow time - (reduces set-up times + wait time (smaller batches))
reduce WIP
reduce market response times

cells = “sociological units conducive to teamwork”

hopefully - motivation to improve ∴ comes naturally


__________________________________________________________________

The cell design process - many aspects - structural + procedural

1. Selection of part families and grouping of parts into families


2. Selection of machine/process populations and grouping of these into
cells
3. Selection of tools, fixtures and pallets
4. Selection of material handling equipment
5. Choice of equipment layout

1. Detailed design of jobs


2. Organization of supervisory and support personnel around the cellular
structure
3. Formulation of maintenance and inspection policies
4. Design of procedures for production planning, scheduling control and
acquisition of related software and hardware
5. Modification of cost control and reward systems
6. Outline of procedures for interfacing with the remaining manufacturing
system (work flow + information, whether computer controlled or not)
Usually - structure oriented ones precede procedural ones

Design will iterate with use


________________________________________________________________

Evaluation of cell design decisions

based either on system structure or system operation

system structure:

1. Equipment and tooling investment (low)


2. Equipment relocation cost (low)
3. Inter- and intracell material handling costs (low)
4. Floor space requirements (low)
5. Extent to which parts are completed in a cell (high)
6. Flexibility (high)

system operation:

1. Equipment utilization (high)


2. Work-in-process inventory (low)
3. Queue lengths at each workstation (short)
4. Job throughput time (short)
5. Job lateness (low)

Typical design - Fanuc cell

cells - evolved from group technology

∴ don’t necessarily need to be


automated.
You are a shop foreman in charge of 5 groups of machines: a set of roughers, a
set of ball millers, a set of finishing mills, a set of drills and a set of lathes.

You have been asked to set up a layout designed for the large scale production
of 10 components which will then be subsequently packaged for shipping.

These components are rather bulky and therefore, movement costs need to be
minimized.

Assuming that the machine layout is to be cellular, then the notional cost of
moving a component within a cell (intracellular) is 1.00 and the cost of moving a
component between cells (intercellular) is 2.00.

Using a notional labelling scheme for the components (i.e. component 1 to


component 10), the following is true:

Components 1 2 3 4 5 7 8 9 10 need to be roughed


Components 2 3 4 9 10 need to be ball milled
Components 1 5 6 7 need to be finished
Components 2 3 4 8 9 10 need to be drilled and
Components 1 2 3 4 5 6 7 8 need to be turned.

Based on this information - come up with a suitable cellular layout with minimal
cost.
To do this - need a cell formation approach

i.e. a mathematical algorithm

For simple examples - manual (visual) approach:

e.g.

4 machines making 6 parts

look for similarities or patterns

clearly - the following groups (cells) should be considered.

However - only possible for v. simple cases

more complex cases - need analytical approach


Rank order clustering (ROC) system

ROC = simple algorithm for producing machine groups (cells)

relies on a machine-part incidence matrix

e.g. our example


__________________________________________________________________
____________

ROC process:

step 1: assign binary weight and calculate a decimal weight for each row
and column using :
m
Decimal weight for row i = ∑b
p =1
ip 2 m− p

n
Decimal weight for colum j = ∑ bpj 2 n− p
p =1

step 2: rank the rows in order of decreasing decimal weight values

step 3: repeat steps 1 and 2 for each column

step 4: continue previous steps until there is no change in the position of


each element in each row and column

step 5: identify groups


Machine-part incidence matrix:

i.e. assign a number to each component and a number to each machine and
tabulate

step 2: assign binary weights to the components and sum

step 3: reorder by ranking, then assign binary weights to machines

reorder according to component sums if necessary, then reorder according to


machine sums if necessary

if no further change – stop

i.e.

note the blocking


process
Single-linkage cluster analysis (SLCA)

Utilizes similarity coefficients

Produces a dendrogram - giving graphical evidence of optimum cell


structures

Similarity coefficient (Sij) between 2 machines (i and j) :


= ratio of no. of parts visiting both machines to the no. of machines
visiting either machine
N

∑X ijk
i.e. Sij = N
k =1

∑ (Y
k =1
ik + Z jk − X ijk )

where Xijk = operation on part k performed both on machine i and j

Yik = operation on part k performed on machine i

Zjk = operation on part k performed on machine j

SCLA algorithm

step 1: compute the similarity coefficients for all possible pairs of


machines

step 2: select the two most similar machines to form the first machine cell

step 3: lower the similarity level and form new machine cells by including
all the machines with similarity coefficients not less than the
threshold value

step 4: continue step 3 until all the machines are grouped into a single cell

step 5: form a dendrogram


using the SCLA algorithm

step 1: e.g. machines 1 and 2:

5
SC12 = = 0.556
9+5−5

all other coefficients found the same way:

Machine pair M1 M1 M1 M1 M2 M2 M2 M3 M3 M4
M2 M3 M4 M5 M3 M4 M5 M4 M5 M5

Similarity coefficient 0.56 0.30 0.67 0.70 0.00 0.83 0.30 0.00 0.50 0.40

Step 2: select machine groups M2 and M4 – the first cell

Step 3: next group = M1/M5 at 0.70 then M1/M4 at 0.67


∴ at a threshold of 0.67 – M1, M2, M4 and M5 form a cell
at a level of 0.5, M3 joins the cell

as a dendrogram:
Evaluation of cell designs

In previous example: no. of cells formed


- dependent on required level of similarity

which one is best?


Depends on cell movements

∴ determine total movement of each component throughout the system

depends if machines are laid out in a line, square or rectangle

3 assumptions:

1. Expected distance a part moves between two machines in a cell of N


machines in a straight line is (N+1)/3
2. Expected distance for a rectangular layout with M rows of L machines is
(M+L)/3

3. Expected distance for a square layout is (2%N)/3

Total distance moved in jth cell for the ith configuration


m
= ∑ d ij kij
j

dij = expected distance moved between two machines for the ith
configuration in the jth cell

kij = number of moves between two machines by all parts for the ith
configuration in the jth cell

For the ith configuration - total movement cost = TCi


m
TCi = C1.N1 + C2 ∑d k
j
ij ij

C1 = cost of an intercell movement


C2 = cost per unit distance of an intracell movement
Ni = number of intercell movements for the ith configuration

Best configuration i is given by the minimum of TCi over all i.


Movement considerations:

consider our solution:

e.g. cell (M1,M5):


= 7 intracellular moves

cell (M2,M4):
= 5 intracellular moves

cell (M3):
= 0 intracellular moves

Total distance moved = 7*(2+1)/3 + 5*(2+1)/3 + 0 = 12 units of distance

Also - 10 intercellular moves

∴ total cost = $2.00*10 + $1.00*12 = $32 - for a 3-cell system

for all systems:

Assume all machines in each cell = in a single line

In 2-cell case - if cell (M1,M2,M4,M5) = 2x2 square

Then total intracellular movement = 24 units - not 30 units

You might also like