MEng6106- Manufacturing Systems Modeling

& Performance Analysis

Lecture II - Assembly Systems and Line

Balancing
Dr. Tafesse Gebresenbet
AAU, Technology Faculty
Mechanical Engineering
Department
Email [email protected]
References : Mikel Groover, Automation, production systems and CIM
Askin, Standridge, Modelling and analysis of Manufacturing
systems
 Assembly Systems and Line Balancing

 Assembly involves the joining together of two or

more separate parts to form a new entity, called a
subassembly, an assembly or some similar name.

 Three major categories of processes used to

accomplish the assembly of the components:

1. Mechanical fastening
2. Joining methods
3. Adhesive bounding

Manufacturing systems modeling &

2 perofrmance analysis (TGS)
 Assembly Systems and Line Balancing

1. Mechanical fastening:A mechanical action

to hold the components together. Includes:
Threaded fasteners: Screw, nuts, bolts, etc.
Very common in industry. Allow to be taken
apart if necessary.
Rivets, crimping, and other methods: he
fastener or one of the components is
mechanically deformed.
Press fits: The two parts are joined together
by pressing one into the other. Once fitted,
the parts are not easily separated.
Snap fits: One or both of the parts elastically
deform when pressed together. Commercial
hardware such as retainers, C-rings, and snap
rings may be used.
3
Sewing and
Manufacturing stitching:
systems modeling & Used to assemble soft,
perofrmance analysis (TGS)
 Assembly Systems and Line Balancing

2. Joining methods: Includes welding, brazing, and

soldering. Molten metal is used to join two or more
components together.
Common feature of welding techniques is that
fusing and melting occur in the metal parts being
joined.
In brazing and soldering, only the filler metal
becomes molten for joining. The metal components
do not melt. Not as strong as welding.

3. Adhesive bonding: Involves the use of an

adhesive material to join components. Two types of
adhesives: thermoplastic and thermosetting.
Thermosetting adhesives are more complicated to
apply, but are
Manufacturing stronger
systems modeling and
& capable of withstanding
4 perofrmance analysis (TGS)
 Assembly systems
The methods used to accomplish assembly processes:
1. Manual single-station assembly: Generally used on
a product that is complex and produced in small
quantities. One or more workers are required depending
on the size of the product. Ex: machine tools, industrial
equipment, aircraft, ships, etc.
2. Manual assembly line: Consist of multiple
workstations. One or more workers perform a portion of
the total assembly work on the product.
3. Automated assembly system: Uses automated
methods at the workstations rather than human beings.
Manual Assembly Lines
 Used in high-production situations where the work can
be divided into small tasks (work elements) and the
tasks assigned to the workstations on the line.
 By giving each worker a limited set of tasks repeatedly,
the worker becomes a specialist in those tasks
and perform more quickly. (Division of labor)
Manufacturing systems modeling &
5 perofrmance analysis (TGS)
 Assembly systems
Transfer of Work Between Workstations
1. Non mechanical Lines: Parts are passed from station to
station by hand. Problems are:
 Starving at stations
 Blocking of stations
As a result, cycle times vary. Buffer stocks are used to
overcome.
2. Moving Conveyor Lines: Use a moving conveyor (ex. A
moving belt, conveyor, etc.) to move the subassemblies
between workstations. The system can be continuous,
intermittent (synchronous), or asynchronous.
Problems of continuously moving conveyor:
Starving
Producing incomplete items
In the moving conveyor line, production rate may be controlled
by means ofq feed rate.
= feed rate
= conveyor speed (feet per minute or meters per second)
= spacing between parts
Manufacturing systems modeling &
6 perofrmance analysis (TGS)
 Assembly systems

Raw work parts are launched onto the line at regular

intervals.
The operator has a certain time period during which
he/she must begin work before the part flows past
the station. This time period is called the tolerance
time.
= tolerance time
= length of the station

Model Variations
It is highly desirable to assign appropriate amount of
work to the stations to equalize the process or
assembly times at the workstations.
This brings thesystems
Manufacturing line modeling
balancing
& problem and the three
7 perofrmance analysis (TGS)
 Assembly systems

1. Single Model Line: Specialized line dedicated to

the production of a single product.
2. Batch-model Line (Multiple parallel lines):
Used for the production of two or more models with
similar sequence of processing or assembly
operations.
3. Mixed-model Line: Several models are
intermixed on the line and are processed
simultaneously.

These cases may be applied to both manual flow

lines and automated flow lines.
Type 2 and 3 are easier to apply to manual flow-lines.
The problem of line balancing becomes more
complicated when going from type 1 to type 3.
Manufacturing systems modeling &
8 perofrmance analysis (TGS)
 Multiple parallel lines
 ADVANTAGES  DISSADVANTAGES
 easy work load balancing  higher setup costs
 increasing scheduling  higher equipment costs
flexibility  higher skill requirements
 job enrichment  slower learning
 higher line availability  complex supervision
 more accountability
As with most problems, multiple objectives exist. By far the
most commonly used objective for analytical models
is minimization of idle time.

However, in practice, real world issues of minimizing

tooling investment, minimizing the maximum lift or strain
by any worker, grouping tasks requiring similar skills,
minimizing movement of existing equipment, and meeting
production targets cannot be overlooked.
Manufacturing systems modeling &
9 perofrmance analysis (TGS)
 Workstation cycle time
 PACED LINES
 Each work station is given exactly the same amount of time to
operate on each unit of product. At the conclusion of this cycle
time TC, the handling system automatically indexes each unit
to the next station
 ROLE OF BUFFERS
 Usually small buffers may be needed in non-automatic
assembly to avoid starving. Without buffers if task times vary,
un paced (asynchronous) lines may be preferable.
 UNPACED LINES (ASYNCHRONOUS)
 The station removes a new unit from the handling system as
soon as it has completed the previous unit, performs the
required tasks, and then forwards the unit on to the next
station.
 PARALLEL WORKSTATIONS IN SERIAL SYSTEMS
 In many serial systems, each station along the line is usually a
Manufacturing systems modeling &
10 perofrmance analysis (TGS)
 The Line Balancing Problems
It is to arrange the individual processing and assembly tasks at
the workstations so that the total time required at each station
is approximately the same.

Very difficult to achieve perfect balance in most practical

situations.
If workstation times are unequal, the slowest station determines
the overall production rate of the line.

TERMINOLOGY
Minimum Rational Work Element. The smallest practical
indivisible tasks into which the job can be divided.

= Time required to carry out this rational work element.

Considered to be constant. In fact it varies in a manual station.
Assumed that they are additive. In fact it changes when two are
Manufacturing systems modeling &
11 perofrmance analysis (TGS)
 Total Work Content. ( ) Sum of the time of all
the work elements to be done on the line.

= Number of work elements that make up the total

work or job.

Workstation Process Time. ( ) The sum of the

times of the work elements done at the station.

n= number of stations

Cycle Time. ( ) Ideal or theoretical cycle time of

the flow line. The time interval between parts
coming off the line.

Manufacturing systems modeling &

12 perofrmance analysis (TGS)
 Balance Delay. (balancing loss) (d) Measure of
the line inefficiency.

The balance delay d will be zero for any values n

and that satisfies the relationship

Minimum number of workstations required to

optimize the balance delay for a specified
may be found by

Manufacturing systems modeling &

13 perofrmance analysis (TGS)
 Line Balancing
The Basic objective of Line Balancing problem
To assign work elements to workstations such
that assembly cost is minimized

Total assembly cost includes:

 Labor cost (while performing tasks)
 Idle time cost
 Focus: minimize idle time
 Limits: production constraints

Manufacturing systems modeling &

14 perofrmance analysis (TGS)
 Line Balancing
Problem formulation
 production rate P (units/time)
 number of parallel lines m
 number of tasks N
 time to perform task i : ti
 total task time T = ∑i=1N ti
 to meet demand: cycle time Tc =m/p
 no worker must be assigned a set of tasks of duration longer
than m/p =Tc
Some Features of the Task
order partially determined
 assembly order constraints IP =(u,v) (i.e. task u must
precede task v)
zoning restrictions
 task pairs to same station ZS
 taskManufacturing
pairs not systems
performed
modelingin
& same workstation ZD
15 perofrmance analysis (TGS)
 Line Balancing

Objective function features

 lowered number stations fill up first
 only stations with at least one task are constructed
 benchmarking gage: proportion of idle time
 idle time = (paid -productive)

BALANCE DELAY
(measures proportion of idle time)
D = (K* Tc -  ti)/(K* Tc)

= idle time/paid time

where K* is the number of stations required by
the solution

Manufacturing systems modeling &

16 perofrmance analysis (TGS)
 Line Balancing
Decision variables
 task i assigned to station k ?
 total number of tasks N
Cost coefficient Cik
NCik≤ Ci, K+1, k =1,2,3, …., K-1
 K is the number of maximum workstations allowed. This allows
forcing tasks onto the lowest numbered stations so that
unused stations may be discarded.
 Xik= {1,0} , 1- if task is assigned to station k; 0- otherwise
 total number of stations k
Problem Formulation
 Minimize  (Cik Xik)
 Subject to:
 ti Xik < Tc (all stations k) [the sum of the tasks assigned doesn’t
exceed cycle time}
 Xik = 1 (all tasks i) [the task is assigned to exactly one
workstation]
Xvh < Xuj (all k) & (u,v) in IP [ adherence to the precedence
restriction]
Manufacturing systems modeling &
17 perofrmance analysis (TGS)
 Line Balancing
Comments
 D is idle time over paid time
 objective does not allocate idle time equally
among stns
 best solutions: good work load balancing
 total task time T =  ti
 Maximum time per station is Tc
 minimum stations (lower bound) Ko = | T/TC |

Manufacturing systems modeling &

18 perofrmance analysis (TGS)
 LINE BALANCING APPROACHES
 Largest Candidate rule
 Kilbridge and Wester’s method
 RPWH
 COMSOAL
 OPTIMAL SOLUTIONS
TREE GENERATION & EXPLORATION
PROBLEM STRUCTURE RULES
FATHOMING RULES

Manufacturing systems modeling &

19 perofrmance analysis (TGS)
 Line Balancing
They are heuristic approaches - based on logic and common sense
rather than on mathematical proof. They do not guarantee an
optimal solution, but result in good solutions which approach the
true optimum.
1. Largest-candidate rule:
PROCEDURE
Step 1:List all elements in descending order of .
Step 2:Start from the top and select an element that
satisfies the precedence requirements and does not cause the
sum of the values at the station to exceed the cycle time .
Step 3:Continue to apply Step 2 until no further elements
can be added without exceeding .
Step 4: Repeat steps 2 and 3 for the other stations until all
the elements have been assigned.
The practical realities of the line balancing problem may not permit
the realization of the most desirable number of stations.
Manufacturing systems modeling &
20 perofrmance analysis (TGS)
 2. Kilbridge and Wester’s method:
PROCEDURE
Step 1: Construct the precedence diagram so that the
nodes with identical precedence are arranged vertically in
columns.
Step 2: List the elements in order of their columns. If
an element can be located in more than one column, list
all the columns by the element to show the transferability
of the element.
Step 3: To assign elements to workstations, start with
the column I elements. Continue to the assignment
procedure in order of column number until the cycle time
is reached. Go on until all elements are allocated.

In general, this method provides a superior line balancing

solution when compared with the largest-candidate rule.
Manufacturing systems modeling &
21 perofrmance analysis (TGS)
 3. Ranked positional weights method:
 Ranked Positional Weight Heuristic
 A single sequence is constructed
 A task is prioritized by cumulative assembly time
associated with itself and its successors
 Tasks are then assigned to the lowest numbered feasible
workstation
PROCEDURE
Step 1: Calculate the ranked positional weight value
(RPW) for each element by summing the element’s Te
together with the Te values for all the elements that
follow it in the arrow chain of the precedence diagram.
Step 2: List the elements in descending order of
their RPW.
Step 3: Assign elements to stations according to
RPW, avoiding precedence constraint and time-cycle
violations.

S(i) successor tasks to task i

Manufacturing systems modeling &
22 perofrmance analysis (TGS)
 Ranked positional weights method
Let tasks be ordered, then Let PW(i) be the positional
weight of taks i and let S(i) be the set of its successors.
Thus
PW(i) = ti + ∑ tj ; j in S(i)

Task r Є S(i) if and only if there is a path of immediate

successor relations from i to r.
Let the Immediate Successors be IS(i) and the Immediate
Predecessors be IP(i)

The RPWH procedure as follows

9. Task ordering: For all i=1,2, …. N compute PW(i) and
order the tasks by increasing values of PW (i)
10. Task assignment: For ranked tasks i, assign them in
sequence to the first feasible station

Manufacturing systems modeling &

23 perofrmance analysis (TGS)
 COMSOAL
4. COMSOAL - A Computerized Line Balancing
Method
PROCEDURE :
Computer Method for Sequencing Operations for
Assembly Lines
 Simple record keeping to allow examination of many
possible sequences
 Sequences are generated by random picking a task
and constructing subsequent tasks
 New stations are opened when needed
 Sequences that exceed the best solution are
discarded
 Better sequences become upper bounds

Step 1: Construct list A, showing all work elements

in one column and the total number of elements that
immediately precede each element in an adjacent
column.
Manufacturing systems modeling &
24 perofrmance analysis (TGS)
 COMSOAL (for generation X trial)
 Array of Number of Immediate Predecessors for each task i NIP(i)
 Array of for which other tasks is i an immediate predecessor WIP(i)
 Array of N tasks TK, c- available time remaining in current workstation
 List of unassigned tasks A
 List of tasks from A with all immediate predecessors assigned B
 List of tasks from B with tasks times not exceeding remaining cycle
time in the current workstation F
1.- SET
x=0, UB=∞, cycle time, c=Tc
2.- Start new sequence:
SET x=x+1, A=TK, NIPW(i) = NIP(i)
3.- Precedence feasibility
FOR all i ЄA, IF NIPW(i) = 0 , ADD i TO B
4. Time feasibility
FOR i ЄB, IF ti< c ADD i TO F .
If F empty , 5 , otherwise 6
5.- Open new station
IDLE=IDLE + c , c = TC
If IDLE > UB , 2, otherwise 3

Manufacturing systems modeling &

25 perofrmance analysis (TGS)
 COMSOAL (contd’)
6.- Select task: SET m = card{F}
Random generate RN Є U(0,1)
LET i* = [m*RN] th TASK from F
Remove i* from A,B,F
c = c - ti
FOR ALL i Є WIP(i*), NIPW=NIPW-1
IF A EMPTY Go to 7, OTHERWISE Go to 3
7.- Schedule completion
IDLE = IDLE + c
IF IDLE < UB , UB = IDLE Go to STORE SCHEDULE
IF x = X , STOP, OTHERWISE Go to 2

Manufacturing systems modeling &

26 perofrmance analysis (TGS)
 Optimization of Line Balancing
Consider the construction of a decision tree with all possible
sequences obeying precedence constraints. The tree starts
at the root and evolves to the first branches; the tasks
without predecessors.
Form there, the next level of possible task evolve. The
process continues until we reach the leaves. The path from
the root to each leaf constitutes a complete sequence.
 The optimal sequence is the one of the leaves but which
one? Optimization procedures are based on searching
decision trees for the optimal leaf.

Manufacturing systems modeling &

27 perofrmance analysis (TGS)
 Optimization of Line Balancing
Depth first Backtracking
In the depth first backtracking we first generate the tree.
Then we start with a leaf and backtrack towards the root
of the tree.

Assume the number of decision to be made is N. The choices

available at stage n depend on earlier decisions.
The tree is generated considering only the ordering of the tasks.
At any stage, eligible tasks are those whose predecessors have
already been included in the partially completed sequence.
Sequences thus created necessarily satisfy precedence
constraints.
One proceeds depth first by growing the tree by
selecting first alternative at each stage until a terminal
leaf is reached. Next we backtrack from the leaf towards
the root until unexplored
Manufacturing branch
systems modeling & is reached. Then we
28 perofrmance analysis (TGS)
 Optimization of Line Balancing
To solve the problem, a sequence is selected and divided
into workstations fulfilling the cycle time constraint and
starting new stations only when absolutely necessary. This
requires the concept of fittable task, i.e., a task if fittable
into a station if
 it fits into the remaining idle time of the station
 It is still unassigned, and
 All its predecessors have been assigned
The procedure is then as follows:
5. Input bound and task data
6. Setup; k=1; p=0; Ck=Tc; B=0
7. Select new task: Find i’= lowest i; i fittabel i>N. Does i’
exists?. If yes got to 6; otherwise go to 3.
8. Assign task: Ai=k; p=p+1; Ck = Ck-ti*; B=0. Is p=N? If
yes got to 6; otherwise go to 3.
9. Open new station: k=k+1; Ck=TC
10.Sequence complete; Save it if best solution
11.Backtrack to B (remove B from station K) If Ck=Tc, k=K-1,
B=Tap; AB=0; Ck-Ck+tB; p=P-1; i*= go to 3.
Manufacturing systems modeling &
29 perofrmance analysis (TGS)
 Optimization of Line Balancing
The problem can be simplified by “pruning” the tree i.e.,
only creating those new nodes required for the search of a
better solution.
Fathoming rules are introduced such that if a node satisfies
one of the rules all paths from that node can be discarded.
This process is called fathoming the node. Once all
remaining nodes are either fathomed or are leaves we are
alone.
The following optimality principles need always be invoked.
 Never close a station while fittabel tasks remain
 If a task makes all others unfittabel, make it a work
station.

The list of fathoming rules is as follows

 Task dominance
 Station dominance
 Solution dominance
Manufacturing systems modeling &
30 perofrmance analysis (TGS)
 Mixed Model lines
The objective function is the same to spread the
work load among stations as evenly as possible
can be expressed as follows. In mixed assembly
model assembly line balancing, total work element
times per shift or per hour are used. and
Minimize (wAT-WL) or minimize

Where w-number of workers or stations ( we are

assuming the Mi =1, so that n=w, w=WL/AT
WL-work load to be accomplished by the workers in
the scheduled time period (min/hr)
AT =available time in the period of interest
(min/hr/worker)
TTs= total service time at station i to perform its
assigned portion of the work
31
 Work load can be calculated

The total time to perform each element in the

work load is calculated

Where TTk– the total time within the workload

that must be allocated to element k for all
products (min)

Total service times at each stations are computed

32
 Measures of balance efficiency for mixed
assembly line balancing corresponds to those in
single model line balancing:

Where Eb- balance efficiency

WL-work load
w – number of workers(stations)
max{TTsi}- maximum value of total service
time among all stations in the solution

Manufacturing systems modeling &

33 perofrmance analysis (TGS)
 SEQUENCING MIXED MODELS
We assume
qj– the proportion of the product type j, j=1,….P to be produced
The first step is to develop an assembly line balance for the
weighted average product.
Let tij be the time to perform the task i on product type j and
sk the set of tasks assigned to the workstation k
We can state an average feasibility condition
∑iЄSk ∑j=1 p qj tij ≤ Tc k=1,….K
In solving this problem we use ti =∑j=1 p qj tij
1.- Initialization: create list of all products to be assigned (A)
2.- Assign a product (List A)
 FOR n from list A, create list B of all product types
assignable without violating constraints
 from list B select product which minimizes the function
| ∑j=1 n ∑iЄSkb ti,j(n) - n Ckb |
 Add product type j* to the nth position
 Remove a product type j* from A IF n < N
 GO TO 1

Manufacturing systems modeling &

34 perofrmance analysis (TGS)
 SEQUENCING MIXED MODELS
Subject to the following constraints
∑j=1 N = Nj j=1,…., p
(to ensure that all items are produced during the
cycle)

nNj – s1 ≤ ∑j=1 n Xjb ≤ nNj/N + s1 n=1,…., N

j= 1, ….P
(to restrict the production rate of each product to be
within s1 of its average rate at all times)

∑b=1 n ∑j=1 P ∑iЄsk tiXjb ≤ (n+s2)Ck n= 1,….N

k=1,….K
(limit the maximum overutilization at all times)
Xjn = 0 or 1
The constraints attempt to restrict unplanned station
Manufacturing systems modeling &
35 perofrmance analysis (TGS)
 UNPACED LINES
 Paced line with K stations and cycle time TC, the
Each time spends KTC in system (throughput
time)
Production rate is 1/ TC
 In a deterministic unpaced line
Production rate is 1/ TC
Time in system is maybe not KTC
 WIP is smaller for unpaced lines

Manufacturing systems modeling &

36 perofrmance analysis (TGS)
 Other Ways to Improve the Line Balance
 Dividing work elements
 Changing work head speeds at automatic stations
 Methods analysis
 Preassembly of components
 Inventory buffers between stations
 Parallel stations

Manufacturing systems modeling &

37 perofrmance analysis (TGS)